diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,157085 @@
+{
+  "best_metric": 0.15624292890437616,
+  "best_model_checkpoint": "xtreme_s_xlsr_300m_voxpopuli_en/checkpoint-25500",
+  "epoch": 10.0,
+  "global_step": 26090,
+  "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": 8.9115,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-07,
+      "loss": 8.6707,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-07,
+      "loss": 8.7698,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-07,
+      "loss": 8.6373,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-07,
+      "loss": 7.8067,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.499999999999999e-07,
+      "loss": 8.2134,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-07,
+      "loss": 8.7439,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.05e-06,
+      "loss": 8.8961,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-06,
+      "loss": 8.814,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3499999999999998e-06,
+      "loss": 8.7208,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4999999999999998e-06,
+      "loss": 8.2712,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6499999999999997e-06,
+      "loss": 7.9486,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-06,
+      "loss": 8.3405,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9499999999999995e-06,
+      "loss": 9.0644,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1e-06,
+      "loss": 8.5629,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2499999999999996e-06,
+      "loss": 8.3004,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-06,
+      "loss": 8.0235,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.55e-06,
+      "loss": 7.9564,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6999999999999996e-06,
+      "loss": 8.3934,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.85e-06,
+      "loss": 8.8725,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 8.5985,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.15e-06,
+      "loss": 8.6717,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2999999999999993e-06,
+      "loss": 8.65,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4499999999999996e-06,
+      "loss": 7.6189,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-06,
+      "loss": 8.1016,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-06,
+      "loss": 9.6676,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7499999999999997e-06,
+      "loss": 8.5248,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.899999999999999e-06,
+      "loss": 8.2622,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.049999999999999e-06,
+      "loss": 8.0915,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-06,
+      "loss": 8.1348,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.35e-06,
+      "loss": 7.9998,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.499999999999999e-06,
+      "loss": 8.6772,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6499999999999995e-06,
+      "loss": 8.3921,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8e-06,
+      "loss": 8.2689,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.95e-06,
+      "loss": 8.5407,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1e-06,
+      "loss": 7.986,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.25e-06,
+      "loss": 8.0715,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 8.0569,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.549999999999999e-06,
+      "loss": 8.5662,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7e-06,
+      "loss": 8.3353,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.85e-06,
+      "loss": 8.1341,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 8.2129,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.1499999999999996e-06,
+      "loss": 7.6661,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.1499999999999996e-06,
+      "loss": 9.5469,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.3e-06,
+      "loss": 8.5542,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.449999999999999e-06,
+      "loss": 9.9633,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 7.9362,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.749999999999999e-06,
+      "loss": 7.8792,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.899999999999999e-06,
+      "loss": 7.7083,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.049999999999999e-06,
+      "loss": 7.3984,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2e-06,
+      "loss": 8.4416,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.35e-06,
+      "loss": 7.9763,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.499999999999999e-06,
+      "loss": 7.7157,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.65e-06,
+      "loss": 7.5642,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 7.0528,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.949999999999998e-06,
+      "loss": 7.0778,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.099999999999999e-06,
+      "loss": 7.758,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.249999999999999e-06,
+      "loss": 7.7208,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.4e-06,
+      "loss": 7.284,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.55e-06,
+      "loss": 6.9602,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.7e-06,
+      "loss": 7.0796,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.849999999999998e-06,
+      "loss": 6.6476,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 7.0339,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.149999999999999e-06,
+      "loss": 7.029,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.299999999999999e-06,
+      "loss": 6.7718,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.45e-06,
+      "loss": 6.6848,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.6e-06,
+      "loss": 6.2513,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.75e-06,
+      "loss": 5.9614,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9e-06,
+      "loss": 6.2393,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.005e-05,
+      "loss": 6.0321,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.02e-05,
+      "loss": 5.7711,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.035e-05,
+      "loss": 5.6492,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.05e-05,
+      "loss": 5.4541,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0649999999999998e-05,
+      "loss": 5.2493,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 4.9625,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0949999999999998e-05,
+      "loss": 5.4962,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1099999999999999e-05,
+      "loss": 5.2089,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1249999999999999e-05,
+      "loss": 5.075,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.14e-05,
+      "loss": 4.872,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.155e-05,
+      "loss": 4.8402,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.17e-05,
+      "loss": 4.4273,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1849999999999998e-05,
+      "loss": 4.7507,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 4.6838,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2149999999999999e-05,
+      "loss": 4.5494,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 4.413,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.245e-05,
+      "loss": 4.3442,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.26e-05,
+      "loss": 4.4134,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.275e-05,
+      "loss": 4.3282,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2899999999999998e-05,
+      "loss": 4.2184,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3049999999999999e-05,
+      "loss": 4.1497,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 4.013,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3349999999999998e-05,
+      "loss": 3.9816,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3499999999999998e-05,
+      "loss": 3.8829,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3649999999999998e-05,
+      "loss": 4.0813,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 3.946,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3949999999999999e-05,
+      "loss": 3.8905,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 3.8444,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4249999999999999e-05,
+      "loss": 3.7576,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.44e-05,
+      "loss": 3.7664,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.455e-05,
+      "loss": 3.7988,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.47e-05,
+      "loss": 3.7415,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.485e-05,
+      "loss": 3.6875,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 3.6417,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5149999999999999e-05,
+      "loss": 3.5968,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.53e-05,
+      "loss": 3.5714,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.545e-05,
+      "loss": 3.5703,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 3.4787,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5749999999999997e-05,
+      "loss": 3.5686,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5899999999999997e-05,
+      "loss": 3.542,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6049999999999997e-05,
+      "loss": 3.4921,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 3.4687,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6349999999999998e-05,
+      "loss": 3.4423,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6499999999999998e-05,
+      "loss": 3.538,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6649999999999998e-05,
+      "loss": 3.4192,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.68e-05,
+      "loss": 3.441,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.695e-05,
+      "loss": 3.4249,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.71e-05,
+      "loss": 3.3949,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.725e-05,
+      "loss": 3.3767,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.74e-05,
+      "loss": 3.3276,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.755e-05,
+      "loss": 3.4292,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7699999999999997e-05,
+      "loss": 3.3789,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7849999999999997e-05,
+      "loss": 3.3584,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 3.3086,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8149999999999997e-05,
+      "loss": 3.3401,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8299999999999998e-05,
+      "loss": 3.3369,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8449999999999998e-05,
+      "loss": 3.3497,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 3.2972,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.875e-05,
+      "loss": 3.3068,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.89e-05,
+      "loss": 3.3103,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.905e-05,
+      "loss": 3.2581,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.92e-05,
+      "loss": 3.2646,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.935e-05,
+      "loss": 3.3004,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.95e-05,
+      "loss": 3.2896,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.965e-05,
+      "loss": 3.2925,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98e-05,
+      "loss": 3.2446,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.995e-05,
+      "loss": 3.248,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.01e-05,
+      "loss": 3.2399,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.025e-05,
+      "loss": 3.2854,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.04e-05,
+      "loss": 3.3552,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.055e-05,
+      "loss": 3.2102,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.07e-05,
+      "loss": 3.2023,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.085e-05,
+      "loss": 3.2154,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1e-05,
+      "loss": 3.199,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1149999999999996e-05,
+      "loss": 3.228,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1299999999999996e-05,
+      "loss": 3.2164,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1449999999999996e-05,
+      "loss": 3.1827,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 3.1914,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1749999999999997e-05,
+      "loss": 3.198,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1899999999999997e-05,
+      "loss": 3.1805,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2049999999999997e-05,
+      "loss": 3.2128,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 3.208,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2349999999999998e-05,
+      "loss": 3.1557,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2499999999999998e-05,
+      "loss": 3.1555,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2649999999999998e-05,
+      "loss": 3.1505,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.28e-05,
+      "loss": 3.1448,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.295e-05,
+      "loss": 3.1361,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.31e-05,
+      "loss": 3.1492,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.325e-05,
+      "loss": 3.1614,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.34e-05,
+      "loss": 3.1267,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3549999999999996e-05,
+      "loss": 3.1318,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3699999999999997e-05,
+      "loss": 3.1126,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3849999999999997e-05,
+      "loss": 3.1383,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 3.1451,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4149999999999997e-05,
+      "loss": 3.1117,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4299999999999998e-05,
+      "loss": 3.1172,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4449999999999998e-05,
+      "loss": 3.0888,
+      "step": 166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 3.0772,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.475e-05,
+      "loss": 3.1037,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.49e-05,
+      "loss": 3.1621,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.505e-05,
+      "loss": 3.1281,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.52e-05,
+      "loss": 3.1597,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.535e-05,
+      "loss": 3.0977,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.55e-05,
+      "loss": 3.0895,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.565e-05,
+      "loss": 3.0809,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 3.0802,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5949999999999997e-05,
+      "loss": 3.1056,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6099999999999997e-05,
+      "loss": 3.0953,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6249999999999998e-05,
+      "loss": 3.073,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 3.0589,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6549999999999995e-05,
+      "loss": 3.0677,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6699999999999995e-05,
+      "loss": 3.0864,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6849999999999995e-05,
+      "loss": 3.0997,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 3.0616,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7149999999999996e-05,
+      "loss": 3.063,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7299999999999996e-05,
+      "loss": 3.0516,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7449999999999996e-05,
+      "loss": 3.0536,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 3.0755,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7749999999999997e-05,
+      "loss": 3.0766,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7899999999999997e-05,
+      "loss": 3.0269,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8049999999999997e-05,
+      "loss": 3.0478,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 3.0303,
+      "step": 191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8349999999999998e-05,
+      "loss": 3.0295,
+      "step": 192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 3.002,
+      "step": 193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.865e-05,
+      "loss": 3.0675,
+      "step": 194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.88e-05,
+      "loss": 3.0248,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.895e-05,
+      "loss": 3.0273,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.91e-05,
+      "loss": 3.0307,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.925e-05,
+      "loss": 3.0197,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.94e-05,
+      "loss": 3.0167,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.955e-05,
+      "loss": 3.0505,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.97e-05,
+      "loss": 3.0038,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.985e-05,
+      "loss": 3.0769,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 3.0067,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0149999999999998e-05,
+      "loss": 3.0136,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0299999999999998e-05,
+      "loss": 2.9874,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0449999999999998e-05,
+      "loss": 3.0193,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.06e-05,
+      "loss": 3.0231,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.0749999999999995e-05,
+      "loss": 3.0088,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.09e-05,
+      "loss": 2.9787,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1049999999999996e-05,
+      "loss": 2.9951,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.119999999999999e-05,
+      "loss": 2.9967,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1349999999999996e-05,
+      "loss": 2.9976,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.149999999999999e-05,
+      "loss": 2.9624,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.165e-05,
+      "loss": 2.9757,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1799999999999994e-05,
+      "loss": 2.971,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.195e-05,
+      "loss": 2.9809,
+      "step": 216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2099999999999994e-05,
+      "loss": 2.9658,
+      "step": 217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.225e-05,
+      "loss": 2.9874,
+      "step": 218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 3.0282,
+      "step": 219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.255e-05,
+      "loss": 2.9616,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2699999999999995e-05,
+      "loss": 2.9582,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.285e-05,
+      "loss": 2.95,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 2.9528,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.315e-05,
+      "loss": 2.9399,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3299999999999996e-05,
+      "loss": 2.9781,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.345e-05,
+      "loss": 2.9811,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.36e-05,
+      "loss": 2.9613,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.375e-05,
+      "loss": 2.9609,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.39e-05,
+      "loss": 2.9537,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.405e-05,
+      "loss": 2.9548,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.42e-05,
+      "loss": 2.9685,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.435e-05,
+      "loss": 2.9444,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.45e-05,
+      "loss": 2.9466,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.465e-05,
+      "loss": 2.9421,
+      "step": 234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.48e-05,
+      "loss": 2.9447,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4949999999999996e-05,
+      "loss": 2.9542,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.51e-05,
+      "loss": 2.9616,
+      "step": 237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5249999999999996e-05,
+      "loss": 2.9553,
+      "step": 238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 2.9489,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.555e-05,
+      "loss": 2.9525,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5699999999999994e-05,
+      "loss": 2.9477,
+      "step": 241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.585e-05,
+      "loss": 2.9374,
+      "step": 242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 2.8967,
+      "step": 243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.615e-05,
+      "loss": 2.9713,
+      "step": 244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6299999999999995e-05,
+      "loss": 2.9237,
+      "step": 245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.645e-05,
+      "loss": 2.9225,
+      "step": 246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.6599999999999995e-05,
+      "loss": 2.9094,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.675e-05,
+      "loss": 2.9274,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.6899999999999996e-05,
+      "loss": 2.9316,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.705e-05,
+      "loss": 2.9001,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.7199999999999996e-05,
+      "loss": 2.9315,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.735e-05,
+      "loss": 2.9217,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.75e-05,
+      "loss": 2.923,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.7649999999999994e-05,
+      "loss": 2.9249,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.78e-05,
+      "loss": 2.8599,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.7949999999999994e-05,
+      "loss": 2.9252,
+      "step": 256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.81e-05,
+      "loss": 2.9584,
+      "step": 257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8249999999999995e-05,
+      "loss": 2.9175,
+      "step": 258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.84e-05,
+      "loss": 2.9235,
+      "step": 259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8549999999999995e-05,
+      "loss": 2.9183,
+      "step": 260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.87e-05,
+      "loss": 2.9248,
+      "step": 261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8849999999999996e-05,
+      "loss": 2.9297,
+      "step": 262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9e-05,
+      "loss": 2.9217,
+      "step": 263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9149999999999996e-05,
+      "loss": 2.9017,
+      "step": 264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.93e-05,
+      "loss": 2.9052,
+      "step": 265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.945e-05,
+      "loss": 2.9269,
+      "step": 266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.96e-05,
+      "loss": 2.9986,
+      "step": 267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.975e-05,
+      "loss": 2.9171,
+      "step": 268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.99e-05,
+      "loss": 2.915,
+      "step": 269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.005e-05,
+      "loss": 2.9072,
+      "step": 270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.02e-05,
+      "loss": 2.9085,
+      "step": 271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.035e-05,
+      "loss": 2.8934,
+      "step": 272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.05e-05,
+      "loss": 2.9218,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.065e-05,
+      "loss": 2.9298,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.08e-05,
+      "loss": 2.953,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.095e-05,
+      "loss": 2.9049,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.11e-05,
+      "loss": 2.905,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.125e-05,
+      "loss": 2.9134,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.14e-05,
+      "loss": 2.8963,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.155e-05,
+      "loss": 2.8804,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.17e-05,
+      "loss": 2.9105,
+      "step": 281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.17e-05,
+      "loss": 4.0951,
+      "step": 282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.185e-05,
+      "loss": 2.8948,
+      "step": 283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.2e-05,
+      "loss": 2.8927,
+      "step": 284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.215e-05,
+      "loss": 2.8921,
+      "step": 285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.229999999999999e-05,
+      "loss": 2.9168,
+      "step": 286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.2449999999999995e-05,
+      "loss": 2.8966,
+      "step": 287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.259999999999999e-05,
+      "loss": 2.9089,
+      "step": 288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.2749999999999996e-05,
+      "loss": 2.8924,
+      "step": 289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.289999999999999e-05,
+      "loss": 2.8906,
+      "step": 290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3049999999999996e-05,
+      "loss": 2.8826,
+      "step": 291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.319999999999999e-05,
+      "loss": 2.8803,
+      "step": 292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.334999999999999e-05,
+      "loss": 2.8876,
+      "step": 293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3499999999999993e-05,
+      "loss": 2.9099,
+      "step": 294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.364999999999999e-05,
+      "loss": 2.8873,
+      "step": 295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3799999999999994e-05,
+      "loss": 2.8722,
+      "step": 296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.394999999999999e-05,
+      "loss": 2.8804,
+      "step": 297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4099999999999995e-05,
+      "loss": 2.8868,
+      "step": 298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.424999999999999e-05,
+      "loss": 2.8921,
+      "step": 299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 2.896,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.454999999999999e-05,
+      "loss": 2.966,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4699999999999996e-05,
+      "loss": 2.8795,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.484999999999999e-05,
+      "loss": 2.8796,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 2.8786,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.514999999999999e-05,
+      "loss": 2.8909,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5299999999999997e-05,
+      "loss": 2.9016,
+      "step": 306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5449999999999993e-05,
+      "loss": 2.8802,
+      "step": 307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.56e-05,
+      "loss": 2.8879,
+      "step": 308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5749999999999994e-05,
+      "loss": 2.8929,
+      "step": 309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.59e-05,
+      "loss": 2.8756,
+      "step": 310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6049999999999994e-05,
+      "loss": 2.8893,
+      "step": 311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.62e-05,
+      "loss": 2.87,
+      "step": 312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6349999999999995e-05,
+      "loss": 2.882,
+      "step": 313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.65e-05,
+      "loss": 2.8746,
+      "step": 314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6649999999999996e-05,
+      "loss": 2.884,
+      "step": 315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.68e-05,
+      "loss": 2.8676,
+      "step": 316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.6949999999999996e-05,
+      "loss": 2.8821,
+      "step": 317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.709999999999999e-05,
+      "loss": 2.8725,
+      "step": 318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7249999999999997e-05,
+      "loss": 2.8861,
+      "step": 319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 2.8627,
+      "step": 320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.755e-05,
+      "loss": 2.8728,
+      "step": 321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7699999999999994e-05,
+      "loss": 2.8384,
+      "step": 322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.785e-05,
+      "loss": 2.8646,
+      "step": 323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 2.8502,
+      "step": 324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.815e-05,
+      "loss": 2.9275,
+      "step": 325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8299999999999995e-05,
+      "loss": 2.8669,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.845e-05,
+      "loss": 2.8648,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8599999999999995e-05,
+      "loss": 2.8722,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.875e-05,
+      "loss": 2.864,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8899999999999996e-05,
+      "loss": 2.8741,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.905e-05,
+      "loss": 2.8715,
+      "step": 331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 2.8932,
+      "step": 332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.935e-05,
+      "loss": 2.8589,
+      "step": 333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.95e-05,
+      "loss": 2.8537,
+      "step": 334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.965e-05,
+      "loss": 2.8613,
+      "step": 335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.98e-05,
+      "loss": 2.8664,
+      "step": 336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.995e-05,
+      "loss": 2.8584,
+      "step": 337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.01e-05,
+      "loss": 2.86,
+      "step": 338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.025e-05,
+      "loss": 2.8651,
+      "step": 339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.04e-05,
+      "loss": 2.8512,
+      "step": 340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.055e-05,
+      "loss": 2.8489,
+      "step": 341
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.07e-05,
+      "loss": 2.8501,
+      "step": 342
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.0849999999999996e-05,
+      "loss": 2.8437,
+      "step": 343
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.1e-05,
+      "loss": 2.8543,
+      "step": 344
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.1149999999999996e-05,
+      "loss": 2.872,
+      "step": 345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.13e-05,
+      "loss": 2.8541,
+      "step": 346
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.145e-05,
+      "loss": 2.8643,
+      "step": 347
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.1599999999999994e-05,
+      "loss": 2.8513,
+      "step": 348
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.174999999999999e-05,
+      "loss": 2.8419,
+      "step": 349
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.1899999999999994e-05,
+      "loss": 2.8296,
+      "step": 350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.204999999999999e-05,
+      "loss": 2.8521,
+      "step": 351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 2.8527,
+      "step": 352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.234999999999999e-05,
+      "loss": 2.8282,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.2499999999999995e-05,
+      "loss": 2.8532,
+      "step": 354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.264999999999999e-05,
+      "loss": 2.8508,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.279999999999999e-05,
+      "loss": 2.8328,
+      "step": 356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.294999999999999e-05,
+      "loss": 2.8522,
+      "step": 357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.309999999999999e-05,
+      "loss": 2.8397,
+      "step": 358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.324999999999999e-05,
+      "loss": 2.8317,
+      "step": 359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 2.8316,
+      "step": 360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.3549999999999994e-05,
+      "loss": 2.8201,
+      "step": 361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.369999999999999e-05,
+      "loss": 2.8237,
+      "step": 362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.3849999999999994e-05,
+      "loss": 2.8266,
+      "step": 363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 2.8408,
+      "step": 364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.4149999999999995e-05,
+      "loss": 2.8381,
+      "step": 365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.429999999999999e-05,
+      "loss": 2.8202,
+      "step": 366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.4449999999999995e-05,
+      "loss": 2.8305,
+      "step": 367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.459999999999999e-05,
+      "loss": 2.8178,
+      "step": 368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.4749999999999996e-05,
+      "loss": 2.8109,
+      "step": 369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.489999999999999e-05,
+      "loss": 2.9185,
+      "step": 370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5049999999999996e-05,
+      "loss": 2.8061,
+      "step": 371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.519999999999999e-05,
+      "loss": 2.8017,
+      "step": 372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.535e-05,
+      "loss": 2.7898,
+      "step": 373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5499999999999994e-05,
+      "loss": 2.7751,
+      "step": 374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.565e-05,
+      "loss": 2.7503,
+      "step": 375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.5799999999999994e-05,
+      "loss": 2.7978,
+      "step": 376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.595e-05,
+      "loss": 2.7695,
+      "step": 377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.6099999999999995e-05,
+      "loss": 2.7972,
+      "step": 378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.625e-05,
+      "loss": 2.7637,
+      "step": 379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 2.7555,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.654999999999999e-05,
+      "loss": 2.7426,
+      "step": 381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.6699999999999996e-05,
+      "loss": 2.7675,
+      "step": 382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.684999999999999e-05,
+      "loss": 2.7216,
+      "step": 383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 2.7071,
+      "step": 384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.714999999999999e-05,
+      "loss": 2.7126,
+      "step": 385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.73e-05,
+      "loss": 2.6765,
+      "step": 386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.7449999999999994e-05,
+      "loss": 2.6575,
+      "step": 387
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.76e-05,
+      "loss": 2.717,
+      "step": 388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.7749999999999994e-05,
+      "loss": 2.666,
+      "step": 389
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.79e-05,
+      "loss": 2.6565,
+      "step": 390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.8049999999999995e-05,
+      "loss": 2.6205,
+      "step": 391
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.82e-05,
+      "loss": 2.6362,
+      "step": 392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.8349999999999995e-05,
+      "loss": 2.578,
+      "step": 393
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.85e-05,
+      "loss": 2.6505,
+      "step": 394
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.8649999999999996e-05,
+      "loss": 2.6123,
+      "step": 395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.88e-05,
+      "loss": 2.6055,
+      "step": 396
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.8949999999999996e-05,
+      "loss": 2.5706,
+      "step": 397
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.91e-05,
+      "loss": 2.576,
+      "step": 398
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.925e-05,
+      "loss": 2.5622,
+      "step": 399
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.94e-05,
+      "loss": 2.6499,
+      "step": 400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.955e-05,
+      "loss": 2.5663,
+      "step": 401
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.97e-05,
+      "loss": 2.517,
+      "step": 402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.985e-05,
+      "loss": 2.4958,
+      "step": 403
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 2.4699,
+      "step": 404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.015e-05,
+      "loss": 2.4543,
+      "step": 405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.0299999999999995e-05,
+      "loss": 2.4655,
+      "step": 406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.045e-05,
+      "loss": 2.538,
+      "step": 407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.0599999999999996e-05,
+      "loss": 2.4575,
+      "step": 408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.075e-05,
+      "loss": 2.434,
+      "step": 409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.0899999999999996e-05,
+      "loss": 2.4813,
+      "step": 410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.104999999999999e-05,
+      "loss": 2.3784,
+      "step": 411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.12e-05,
+      "loss": 2.3753,
+      "step": 412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.134999999999999e-05,
+      "loss": 2.4321,
+      "step": 413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.149999999999999e-05,
+      "loss": 2.4669,
+      "step": 414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.165e-05,
+      "loss": 2.3138,
+      "step": 415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.18e-05,
+      "loss": 2.3095,
+      "step": 416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.194999999999999e-05,
+      "loss": 2.3333,
+      "step": 417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.209999999999999e-05,
+      "loss": 2.3181,
+      "step": 418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.225e-05,
+      "loss": 2.3906,
+      "step": 419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 2.274,
+      "step": 420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.254999999999999e-05,
+      "loss": 2.2337,
+      "step": 421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.269999999999999e-05,
+      "loss": 2.2767,
+      "step": 422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.285e-05,
+      "loss": 2.2654,
+      "step": 423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 2.2037,
+      "step": 424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.314999999999999e-05,
+      "loss": 2.2627,
+      "step": 425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.33e-05,
+      "loss": 2.2044,
+      "step": 426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.345e-05,
+      "loss": 2.1806,
+      "step": 427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.359999999999999e-05,
+      "loss": 2.1356,
+      "step": 428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.374999999999999e-05,
+      "loss": 2.1719,
+      "step": 429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.39e-05,
+      "loss": 2.1187,
+      "step": 430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.405e-05,
+      "loss": 2.1632,
+      "step": 431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.419999999999999e-05,
+      "loss": 2.202,
+      "step": 432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.434999999999999e-05,
+      "loss": 2.0622,
+      "step": 433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.45e-05,
+      "loss": 2.043,
+      "step": 434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.465e-05,
+      "loss": 2.064,
+      "step": 435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.479999999999999e-05,
+      "loss": 2.0085,
+      "step": 436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.494999999999999e-05,
+      "loss": 2.0228,
+      "step": 437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.51e-05,
+      "loss": 2.1,
+      "step": 438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.525e-05,
+      "loss": 1.9489,
+      "step": 439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 1.9267,
+      "step": 440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.555e-05,
+      "loss": 1.9486,
+      "step": 441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.57e-05,
+      "loss": 1.9182,
+      "step": 442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.584999999999999e-05,
+      "loss": 1.8937,
+      "step": 443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 1.9816,
+      "step": 444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.615e-05,
+      "loss": 1.9197,
+      "step": 445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.63e-05,
+      "loss": 1.8102,
+      "step": 446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.644999999999999e-05,
+      "loss": 1.8475,
+      "step": 447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.659999999999999e-05,
+      "loss": 1.8385,
+      "step": 448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.675e-05,
+      "loss": 1.8596,
+      "step": 449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.69e-05,
+      "loss": 2.024,
+      "step": 450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.704999999999999e-05,
+      "loss": 2.0323,
+      "step": 451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.72e-05,
+      "loss": 1.9297,
+      "step": 452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.735e-05,
+      "loss": 1.7319,
+      "step": 453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.75e-05,
+      "loss": 1.8525,
+      "step": 454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.764999999999999e-05,
+      "loss": 1.8493,
+      "step": 455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.78e-05,
+      "loss": 1.8198,
+      "step": 456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.795e-05,
+      "loss": 2.0007,
+      "step": 457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.81e-05,
+      "loss": 1.7207,
+      "step": 458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.824999999999999e-05,
+      "loss": 1.6263,
+      "step": 459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.84e-05,
+      "loss": 1.6786,
+      "step": 460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.855e-05,
+      "loss": 1.7486,
+      "step": 461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.87e-05,
+      "loss": 1.7671,
+      "step": 462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.884999999999999e-05,
+      "loss": 1.755,
+      "step": 463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.9e-05,
+      "loss": 1.6405,
+      "step": 464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.915e-05,
+      "loss": 1.5943,
+      "step": 465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.93e-05,
+      "loss": 1.5964,
+      "step": 466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.945e-05,
+      "loss": 1.5702,
+      "step": 467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.96e-05,
+      "loss": 1.595,
+      "step": 468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.975e-05,
+      "loss": 1.7659,
+      "step": 469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.989999999999999e-05,
+      "loss": 1.509,
+      "step": 470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.005e-05,
+      "loss": 1.5558,
+      "step": 471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.02e-05,
+      "loss": 1.4631,
+      "step": 472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.034999999999999e-05,
+      "loss": 1.5748,
+      "step": 473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.049999999999999e-05,
+      "loss": 1.5368,
+      "step": 474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.065e-05,
+      "loss": 1.7712,
+      "step": 475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.079999999999999e-05,
+      "loss": 1.4881,
+      "step": 476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.094999999999999e-05,
+      "loss": 1.4378,
+      "step": 477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.11e-05,
+      "loss": 1.4659,
+      "step": 478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.125e-05,
+      "loss": 1.398,
+      "step": 479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 1.4754,
+      "step": 480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.154999999999999e-05,
+      "loss": 1.4916,
+      "step": 481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.17e-05,
+      "loss": 1.5459,
+      "step": 482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.184999999999998e-05,
+      "loss": 1.4278,
+      "step": 483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 1.404,
+      "step": 484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.214999999999999e-05,
+      "loss": 1.3232,
+      "step": 485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.23e-05,
+      "loss": 1.3691,
+      "step": 486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.244999999999999e-05,
+      "loss": 1.4755,
+      "step": 487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.259999999999999e-05,
+      "loss": 1.5269,
+      "step": 488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.274999999999999e-05,
+      "loss": 1.4298,
+      "step": 489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.29e-05,
+      "loss": 1.3326,
+      "step": 490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.304999999999999e-05,
+      "loss": 1.3621,
+      "step": 491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.319999999999999e-05,
+      "loss": 1.3652,
+      "step": 492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.335e-05,
+      "loss": 1.3627,
+      "step": 493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.35e-05,
+      "loss": 1.5318,
+      "step": 494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.364999999999999e-05,
+      "loss": 1.3409,
+      "step": 495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.379999999999999e-05,
+      "loss": 1.2823,
+      "step": 496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.395e-05,
+      "loss": 1.2622,
+      "step": 497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.41e-05,
+      "loss": 1.3304,
+      "step": 498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.424999999999999e-05,
+      "loss": 1.2719,
+      "step": 499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 1.4221,
+      "step": 500
+    },
+    {
+      "epoch": 0.19,
+      "eval_cer": 0.3431726398345592,
+      "eval_loss": 1.3325395584106445,
+      "eval_runtime": 46.9595,
+      "eval_samples_per_second": 37.33,
+      "eval_steps_per_second": 4.685,
+      "eval_wer": 0.822396705435127,
+      "step": 500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.455e-05,
+      "loss": 1.2771,
+      "step": 501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.47e-05,
+      "loss": 1.341,
+      "step": 502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.484999999999999e-05,
+      "loss": 1.2805,
+      "step": 503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.5e-05,
+      "loss": 1.2835,
+      "step": 504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.515e-05,
+      "loss": 1.2803,
+      "step": 505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.529999999999999e-05,
+      "loss": 1.3385,
+      "step": 506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.544999999999999e-05,
+      "loss": 1.3736,
+      "step": 507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.56e-05,
+      "loss": 1.1878,
+      "step": 508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.575e-05,
+      "loss": 1.2716,
+      "step": 509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.589999999999999e-05,
+      "loss": 1.2471,
+      "step": 510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.604999999999999e-05,
+      "loss": 1.1523,
+      "step": 511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.62e-05,
+      "loss": 1.2002,
+      "step": 512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.635e-05,
+      "loss": 1.436,
+      "step": 513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.649999999999999e-05,
+      "loss": 1.2022,
+      "step": 514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.664999999999999e-05,
+      "loss": 1.3179,
+      "step": 515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.68e-05,
+      "loss": 1.2196,
+      "step": 516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.695e-05,
+      "loss": 1.2009,
+      "step": 517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.709999999999999e-05,
+      "loss": 1.2579,
+      "step": 518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.725e-05,
+      "loss": 1.5469,
+      "step": 519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.74e-05,
+      "loss": 1.1638,
+      "step": 520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.755e-05,
+      "loss": 1.158,
+      "step": 521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.769999999999999e-05,
+      "loss": 1.237,
+      "step": 522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.785e-05,
+      "loss": 1.212,
+      "step": 523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.8e-05,
+      "loss": 1.2883,
+      "step": 524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.815e-05,
+      "loss": 1.6281,
+      "step": 525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.829999999999999e-05,
+      "loss": 1.1816,
+      "step": 526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.845e-05,
+      "loss": 1.0942,
+      "step": 527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.86e-05,
+      "loss": 1.2118,
+      "step": 528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.874999999999999e-05,
+      "loss": 1.1586,
+      "step": 529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.89e-05,
+      "loss": 1.1365,
+      "step": 530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.905e-05,
+      "loss": 1.3815,
+      "step": 531
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.92e-05,
+      "loss": 1.3026,
+      "step": 532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.934999999999999e-05,
+      "loss": 1.1855,
+      "step": 533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.95e-05,
+      "loss": 1.2276,
+      "step": 534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.965e-05,
+      "loss": 1.1298,
+      "step": 535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.98e-05,
+      "loss": 1.2212,
+      "step": 536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.994999999999999e-05,
+      "loss": 1.2488,
+      "step": 537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.01e-05,
+      "loss": 1.4869,
+      "step": 538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.025e-05,
+      "loss": 1.1076,
+      "step": 539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.04e-05,
+      "loss": 1.239,
+      "step": 540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.054999999999999e-05,
+      "loss": 1.1013,
+      "step": 541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.07e-05,
+      "loss": 1.0631,
+      "step": 542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.085e-05,
+      "loss": 1.1721,
+      "step": 543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.1e-05,
+      "loss": 1.3731,
+      "step": 544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.115e-05,
+      "loss": 1.1095,
+      "step": 545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.13e-05,
+      "loss": 1.1915,
+      "step": 546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.145e-05,
+      "loss": 1.0592,
+      "step": 547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.16e-05,
+      "loss": 1.0189,
+      "step": 548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.175e-05,
+      "loss": 1.0438,
+      "step": 549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.19e-05,
+      "loss": 1.4108,
+      "step": 550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.205e-05,
+      "loss": 1.0677,
+      "step": 551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.22e-05,
+      "loss": 1.4867,
+      "step": 552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.235e-05,
+      "loss": 1.1907,
+      "step": 553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.25e-05,
+      "loss": 1.1749,
+      "step": 554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.265e-05,
+      "loss": 1.1286,
+      "step": 555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.28e-05,
+      "loss": 1.3951,
+      "step": 556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.295e-05,
+      "loss": 1.2965,
+      "step": 557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.31e-05,
+      "loss": 1.0653,
+      "step": 558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.325e-05,
+      "loss": 1.1262,
+      "step": 559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.34e-05,
+      "loss": 0.9737,
+      "step": 560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.355e-05,
+      "loss": 1.1976,
+      "step": 561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.37e-05,
+      "loss": 1.1886,
+      "step": 562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.385e-05,
+      "loss": 1.369,
+      "step": 563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.4e-05,
+      "loss": 1.1196,
+      "step": 564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.415e-05,
+      "loss": 1.0572,
+      "step": 565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.43e-05,
+      "loss": 1.0599,
+      "step": 566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.444999999999998e-05,
+      "loss": 1.0804,
+      "step": 567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.459999999999998e-05,
+      "loss": 1.1462,
+      "step": 568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.474999999999999e-05,
+      "loss": 1.3575,
+      "step": 569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.489999999999999e-05,
+      "loss": 1.078,
+      "step": 570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.504999999999998e-05,
+      "loss": 1.08,
+      "step": 571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.519999999999998e-05,
+      "loss": 1.0828,
+      "step": 572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.534999999999999e-05,
+      "loss": 1.0559,
+      "step": 573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.549999999999999e-05,
+      "loss": 1.0657,
+      "step": 574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.564999999999998e-05,
+      "loss": 1.3658,
+      "step": 575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.579999999999998e-05,
+      "loss": 1.3853,
+      "step": 576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.594999999999999e-05,
+      "loss": 1.0679,
+      "step": 577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.609999999999999e-05,
+      "loss": 1.1357,
+      "step": 578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.624999999999998e-05,
+      "loss": 1.0704,
+      "step": 579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.639999999999999e-05,
+      "loss": 1.0342,
+      "step": 580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.654999999999999e-05,
+      "loss": 1.2662,
+      "step": 581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.669999999999998e-05,
+      "loss": 1.4046,
+      "step": 582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.684999999999998e-05,
+      "loss": 1.0566,
+      "step": 583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 0.9689,
+      "step": 584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.714999999999999e-05,
+      "loss": 1.032,
+      "step": 585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.729999999999998e-05,
+      "loss": 1.0247,
+      "step": 586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.744999999999998e-05,
+      "loss": 1.0967,
+      "step": 587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.759999999999999e-05,
+      "loss": 1.4216,
+      "step": 588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.774999999999999e-05,
+      "loss": 0.9841,
+      "step": 589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.789999999999998e-05,
+      "loss": 1.0339,
+      "step": 590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.804999999999999e-05,
+      "loss": 1.0019,
+      "step": 591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.819999999999999e-05,
+      "loss": 0.9976,
+      "step": 592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.834999999999999e-05,
+      "loss": 1.1059,
+      "step": 593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.849999999999998e-05,
+      "loss": 1.2472,
+      "step": 594
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.864999999999999e-05,
+      "loss": 0.9754,
+      "step": 595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.879999999999999e-05,
+      "loss": 1.042,
+      "step": 596
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.895e-05,
+      "loss": 1.0628,
+      "step": 597
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.909999999999998e-05,
+      "loss": 1.0149,
+      "step": 598
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.924999999999999e-05,
+      "loss": 0.993,
+      "step": 599
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 1.3807,
+      "step": 600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955e-05,
+      "loss": 1.0093,
+      "step": 601
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969999999999998e-05,
+      "loss": 0.9415,
+      "step": 602
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984999999999999e-05,
+      "loss": 0.969,
+      "step": 603
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 1.0252,
+      "step": 604
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.014999999999998e-05,
+      "loss": 1.0037,
+      "step": 605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029999999999999e-05,
+      "loss": 1.0941,
+      "step": 606
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.044999999999999e-05,
+      "loss": 1.1303,
+      "step": 607
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.059999999999999e-05,
+      "loss": 1.062,
+      "step": 608
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.074999999999998e-05,
+      "loss": 1.045,
+      "step": 609
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.089999999999999e-05,
+      "loss": 0.9465,
+      "step": 610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.104999999999999e-05,
+      "loss": 0.97,
+      "step": 611
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.12e-05,
+      "loss": 0.9464,
+      "step": 612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.134999999999998e-05,
+      "loss": 1.1554,
+      "step": 613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.149999999999999e-05,
+      "loss": 1.0508,
+      "step": 614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.164999999999999e-05,
+      "loss": 0.9844,
+      "step": 615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.18e-05,
+      "loss": 0.9409,
+      "step": 616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.194999999999999e-05,
+      "loss": 1.0145,
+      "step": 617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.209999999999999e-05,
+      "loss": 1.0344,
+      "step": 618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.224999999999999e-05,
+      "loss": 1.1982,
+      "step": 619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.24e-05,
+      "loss": 1.0011,
+      "step": 620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.254999999999999e-05,
+      "loss": 0.9691,
+      "step": 621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.269999999999999e-05,
+      "loss": 0.9865,
+      "step": 622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.285e-05,
+      "loss": 0.9806,
+      "step": 623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.3e-05,
+      "loss": 1.0197,
+      "step": 624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.314999999999999e-05,
+      "loss": 1.3423,
+      "step": 625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.329999999999999e-05,
+      "loss": 0.9823,
+      "step": 626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.345e-05,
+      "loss": 0.9478,
+      "step": 627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.36e-05,
+      "loss": 0.8976,
+      "step": 628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.374999999999999e-05,
+      "loss": 0.9479,
+      "step": 629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.389999999999999e-05,
+      "loss": 0.9877,
+      "step": 630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.405e-05,
+      "loss": 1.1417,
+      "step": 631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.419999999999999e-05,
+      "loss": 1.197,
+      "step": 632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.434999999999999e-05,
+      "loss": 1.0326,
+      "step": 633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.449999999999999e-05,
+      "loss": 0.906,
+      "step": 634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.465e-05,
+      "loss": 0.8213,
+      "step": 635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.479999999999999e-05,
+      "loss": 1.1716,
+      "step": 636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.494999999999999e-05,
+      "loss": 1.135,
+      "step": 637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.51e-05,
+      "loss": 1.2504,
+      "step": 638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.525e-05,
+      "loss": 0.9369,
+      "step": 639
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.539999999999999e-05,
+      "loss": 1.0885,
+      "step": 640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.554999999999999e-05,
+      "loss": 0.866,
+      "step": 641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.57e-05,
+      "loss": 0.9432,
+      "step": 642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.585e-05,
+      "loss": 0.9573,
+      "step": 643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 1.2164,
+      "step": 644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.614999999999999e-05,
+      "loss": 0.9894,
+      "step": 645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.63e-05,
+      "loss": 0.9153,
+      "step": 646
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.645e-05,
+      "loss": 0.8798,
+      "step": 647
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.659999999999999e-05,
+      "loss": 0.938,
+      "step": 648
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.675e-05,
+      "loss": 0.9283,
+      "step": 649
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.69e-05,
+      "loss": 1.2918,
+      "step": 650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.705e-05,
+      "loss": 1.4366,
+      "step": 651
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.719999999999999e-05,
+      "loss": 0.9324,
+      "step": 652
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.735e-05,
+      "loss": 0.9516,
+      "step": 653
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.75e-05,
+      "loss": 0.9362,
+      "step": 654
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.764999999999999e-05,
+      "loss": 0.9116,
+      "step": 655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.779999999999999e-05,
+      "loss": 1.178,
+      "step": 656
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.795e-05,
+      "loss": 1.0919,
+      "step": 657
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.81e-05,
+      "loss": 0.9466,
+      "step": 658
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.824999999999999e-05,
+      "loss": 0.8822,
+      "step": 659
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.839999999999999e-05,
+      "loss": 0.8769,
+      "step": 660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.855e-05,
+      "loss": 0.8529,
+      "step": 661
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.87e-05,
+      "loss": 0.9159,
+      "step": 662
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.884999999999999e-05,
+      "loss": 1.0948,
+      "step": 663
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.9e-05,
+      "loss": 0.8507,
+      "step": 664
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.915e-05,
+      "loss": 0.7981,
+      "step": 665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.93e-05,
+      "loss": 0.8291,
+      "step": 666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.944999999999999e-05,
+      "loss": 0.8555,
+      "step": 667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.96e-05,
+      "loss": 1.0854,
+      "step": 668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.975e-05,
+      "loss": 1.1126,
+      "step": 669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.99e-05,
+      "loss": 0.851,
+      "step": 670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010004999999999999,
+      "loss": 0.8534,
+      "step": 671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001002,
+      "loss": 0.8706,
+      "step": 672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010035,
+      "loss": 0.9107,
+      "step": 673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001005,
+      "loss": 0.8772,
+      "step": 674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010065,
+      "loss": 1.2078,
+      "step": 675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001008,
+      "loss": 0.8946,
+      "step": 676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010095,
+      "loss": 0.895,
+      "step": 677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001011,
+      "loss": 1.0218,
+      "step": 678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010125,
+      "loss": 0.8596,
+      "step": 679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001014,
+      "loss": 0.8246,
+      "step": 680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010155,
+      "loss": 1.0716,
+      "step": 681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010169999999999999,
+      "loss": 0.9685,
+      "step": 682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010185,
+      "loss": 0.8789,
+      "step": 683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000102,
+      "loss": 0.8648,
+      "step": 684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010215,
+      "loss": 0.877,
+      "step": 685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010229999999999999,
+      "loss": 0.8084,
+      "step": 686
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010245,
+      "loss": 1.0347,
+      "step": 687
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001026,
+      "loss": 1.0815,
+      "step": 688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010275,
+      "loss": 0.8691,
+      "step": 689
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001029,
+      "loss": 0.9002,
+      "step": 690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00010305,
+      "loss": 0.8978,
+      "step": 691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010319999999999999,
+      "loss": 0.845,
+      "step": 692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010334999999999998,
+      "loss": 1.0279,
+      "step": 693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010349999999999998,
+      "loss": 1.063,
+      "step": 694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010364999999999999,
+      "loss": 0.8655,
+      "step": 695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010379999999999999,
+      "loss": 0.8451,
+      "step": 696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010394999999999998,
+      "loss": 0.8298,
+      "step": 697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010409999999999998,
+      "loss": 0.9474,
+      "step": 698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010424999999999999,
+      "loss": 0.7902,
+      "step": 699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010439999999999999,
+      "loss": 1.1485,
+      "step": 700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010454999999999998,
+      "loss": 0.8321,
+      "step": 701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010469999999999998,
+      "loss": 1.0721,
+      "step": 702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010484999999999999,
+      "loss": 0.8237,
+      "step": 703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 0.787,
+      "step": 704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010514999999999998,
+      "loss": 0.9215,
+      "step": 705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010529999999999998,
+      "loss": 1.0113,
+      "step": 706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010544999999999999,
+      "loss": 1.0664,
+      "step": 707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 0.9451,
+      "step": 708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010574999999999998,
+      "loss": 0.8836,
+      "step": 709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010589999999999999,
+      "loss": 0.7914,
+      "step": 710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010604999999999999,
+      "loss": 0.8737,
+      "step": 711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010619999999999998,
+      "loss": 1.0078,
+      "step": 712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010634999999999998,
+      "loss": 1.0565,
+      "step": 713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010649999999999999,
+      "loss": 0.8378,
+      "step": 714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010664999999999999,
+      "loss": 0.9146,
+      "step": 715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010679999999999998,
+      "loss": 0.8578,
+      "step": 716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00010694999999999998,
+      "loss": 0.8836,
+      "step": 717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010709999999999999,
+      "loss": 0.8511,
+      "step": 718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010724999999999999,
+      "loss": 1.0557,
+      "step": 719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010739999999999998,
+      "loss": 0.808,
+      "step": 720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010754999999999999,
+      "loss": 0.9053,
+      "step": 721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010769999999999999,
+      "loss": 0.8057,
+      "step": 722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010784999999999999,
+      "loss": 0.778,
+      "step": 723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 0.8308,
+      "step": 724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010814999999999999,
+      "loss": 1.2852,
+      "step": 725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010829999999999999,
+      "loss": 1.0224,
+      "step": 726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010845,
+      "loss": 0.7915,
+      "step": 727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010859999999999998,
+      "loss": 0.8817,
+      "step": 728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010874999999999999,
+      "loss": 0.8626,
+      "step": 729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010889999999999999,
+      "loss": 0.7672,
+      "step": 730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010904999999999998,
+      "loss": 0.9232,
+      "step": 731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010919999999999998,
+      "loss": 0.9226,
+      "step": 732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010934999999999999,
+      "loss": 0.8089,
+      "step": 733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010949999999999999,
+      "loss": 0.7869,
+      "step": 734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010964999999999998,
+      "loss": 0.9092,
+      "step": 735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010979999999999999,
+      "loss": 0.8277,
+      "step": 736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00010994999999999999,
+      "loss": 0.8518,
+      "step": 737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011009999999999999,
+      "loss": 0.9794,
+      "step": 738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011024999999999998,
+      "loss": 0.7762,
+      "step": 739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011039999999999999,
+      "loss": 1.181,
+      "step": 740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011054999999999999,
+      "loss": 0.909,
+      "step": 741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001107,
+      "loss": 0.8491,
+      "step": 742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011084999999999998,
+      "loss": 1.0055,
+      "step": 743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 1.156,
+      "step": 744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011114999999999999,
+      "loss": 0.7996,
+      "step": 745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001113,
+      "loss": 1.4307,
+      "step": 746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011144999999999998,
+      "loss": 0.8358,
+      "step": 747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011159999999999999,
+      "loss": 0.9697,
+      "step": 748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011174999999999999,
+      "loss": 0.9706,
+      "step": 749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001119,
+      "loss": 1.2269,
+      "step": 750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011204999999999999,
+      "loss": 0.9103,
+      "step": 751
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011219999999999999,
+      "loss": 0.8737,
+      "step": 752
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011235,
+      "loss": 0.7994,
+      "step": 753
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001125,
+      "loss": 0.7954,
+      "step": 754
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011264999999999999,
+      "loss": 1.0558,
+      "step": 755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 1.0079,
+      "step": 756
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011295,
+      "loss": 1.0779,
+      "step": 757
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011309999999999998,
+      "loss": 0.8591,
+      "step": 758
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011324999999999999,
+      "loss": 0.7826,
+      "step": 759
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 0.8095,
+      "step": 760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011355,
+      "loss": 0.8423,
+      "step": 761
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011369999999999999,
+      "loss": 0.8628,
+      "step": 762
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011384999999999999,
+      "loss": 1.0739,
+      "step": 763
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 0.7815,
+      "step": 764
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011415,
+      "loss": 0.8404,
+      "step": 765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011429999999999999,
+      "loss": 0.7889,
+      "step": 766
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011444999999999999,
+      "loss": 0.8712,
+      "step": 767
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001146,
+      "loss": 0.8008,
+      "step": 768
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011475,
+      "loss": 1.0067,
+      "step": 769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011489999999999999,
+      "loss": 0.773,
+      "step": 770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011504999999999999,
+      "loss": 0.7613,
+      "step": 771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001152,
+      "loss": 0.8078,
+      "step": 772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011535,
+      "loss": 0.814,
+      "step": 773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011549999999999999,
+      "loss": 0.7906,
+      "step": 774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011564999999999999,
+      "loss": 1.0821,
+      "step": 775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001158,
+      "loss": 0.7594,
+      "step": 776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011595,
+      "loss": 0.8595,
+      "step": 777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011609999999999999,
+      "loss": 0.7957,
+      "step": 778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011624999999999999,
+      "loss": 0.8365,
+      "step": 779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001164,
+      "loss": 0.7618,
+      "step": 780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011654999999999999,
+      "loss": 0.8725,
+      "step": 781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011669999999999999,
+      "loss": 0.975,
+      "step": 782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011685,
+      "loss": 0.8661,
+      "step": 783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000117,
+      "loss": 0.7799,
+      "step": 784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011714999999999999,
+      "loss": 0.6903,
+      "step": 785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011729999999999999,
+      "loss": 0.7141,
+      "step": 786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011745,
+      "loss": 0.8898,
+      "step": 787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001176,
+      "loss": 1.1432,
+      "step": 788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011774999999999999,
+      "loss": 0.7109,
+      "step": 789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011789999999999999,
+      "loss": 0.8003,
+      "step": 790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011805,
+      "loss": 0.7553,
+      "step": 791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001182,
+      "loss": 0.9,
+      "step": 792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011834999999999999,
+      "loss": 0.8573,
+      "step": 793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001185,
+      "loss": 0.9733,
+      "step": 794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011865,
+      "loss": 0.7305,
+      "step": 795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001188,
+      "loss": 0.7742,
+      "step": 796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011894999999999999,
+      "loss": 0.7168,
+      "step": 797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001191,
+      "loss": 0.6995,
+      "step": 798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011925,
+      "loss": 0.8462,
+      "step": 799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001194,
+      "loss": 1.0784,
+      "step": 800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011954999999999999,
+      "loss": 0.8028,
+      "step": 801
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001197,
+      "loss": 0.706,
+      "step": 802
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011985,
+      "loss": 0.7485,
+      "step": 803
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 0.7786,
+      "step": 804
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00012014999999999999,
+      "loss": 0.716,
+      "step": 805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001203,
+      "loss": 0.9206,
+      "step": 806
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00012045,
+      "loss": 0.9471,
+      "step": 807
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00012059999999999999,
+      "loss": 0.7563,
+      "step": 808
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00012075,
+      "loss": 0.7352,
+      "step": 809
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001209,
+      "loss": 0.721,
+      "step": 810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00012105,
+      "loss": 0.7461,
+      "step": 811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00012119999999999999,
+      "loss": 0.9241,
+      "step": 812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00012135,
+      "loss": 0.8893,
+      "step": 813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001215,
+      "loss": 0.8197,
+      "step": 814
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00012165,
+      "loss": 0.7652,
+      "step": 815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00012179999999999999,
+      "loss": 0.9016,
+      "step": 816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00012194999999999998,
+      "loss": 0.7514,
+      "step": 817
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00012209999999999999,
+      "loss": 0.7667,
+      "step": 818
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00012225,
+      "loss": 1.1199,
+      "step": 819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001224,
+      "loss": 0.8649,
+      "step": 820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00012254999999999997,
+      "loss": 0.6772,
+      "step": 821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012269999999999997,
+      "loss": 0.7754,
+      "step": 822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012284999999999998,
+      "loss": 1.0186,
+      "step": 823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 0.7309,
+      "step": 824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012314999999999998,
+      "loss": 1.1697,
+      "step": 825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001233,
+      "loss": 1.1182,
+      "step": 826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012345,
+      "loss": 0.7119,
+      "step": 827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001236,
+      "loss": 0.7034,
+      "step": 828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012374999999999997,
+      "loss": 0.7055,
+      "step": 829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012389999999999998,
+      "loss": 0.8283,
+      "step": 830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012404999999999998,
+      "loss": 0.7282,
+      "step": 831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012419999999999998,
+      "loss": 0.8532,
+      "step": 832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012435,
+      "loss": 0.7274,
+      "step": 833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001245,
+      "loss": 0.7238,
+      "step": 834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012465,
+      "loss": 0.7889,
+      "step": 835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012479999999999997,
+      "loss": 0.7645,
+      "step": 836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012494999999999997,
+      "loss": 0.8967,
+      "step": 837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012509999999999998,
+      "loss": 1.129,
+      "step": 838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012524999999999998,
+      "loss": 0.7055,
+      "step": 839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012539999999999999,
+      "loss": 0.8423,
+      "step": 840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012555,
+      "loss": 0.6795,
+      "step": 841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001257,
+      "loss": 0.8369,
+      "step": 842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012585,
+      "loss": 0.8985,
+      "step": 843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 1.0388,
+      "step": 844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012614999999999998,
+      "loss": 0.7367,
+      "step": 845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012629999999999998,
+      "loss": 0.8164,
+      "step": 846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00012644999999999998,
+      "loss": 0.7395,
+      "step": 847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001266,
+      "loss": 0.8355,
+      "step": 848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012675,
+      "loss": 0.7598,
+      "step": 849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001269,
+      "loss": 1.0105,
+      "step": 850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012705,
+      "loss": 0.8749,
+      "step": 851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012719999999999997,
+      "loss": 0.7161,
+      "step": 852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012734999999999998,
+      "loss": 0.7503,
+      "step": 853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012749999999999998,
+      "loss": 0.6659,
+      "step": 854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012764999999999999,
+      "loss": 0.7271,
+      "step": 855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001278,
+      "loss": 0.9715,
+      "step": 856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012795,
+      "loss": 1.0386,
+      "step": 857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001281,
+      "loss": 0.74,
+      "step": 858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012824999999999997,
+      "loss": 0.734,
+      "step": 859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012839999999999998,
+      "loss": 0.6723,
+      "step": 860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012854999999999998,
+      "loss": 0.6255,
+      "step": 861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012869999999999998,
+      "loss": 0.9069,
+      "step": 862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012885,
+      "loss": 0.9487,
+      "step": 863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000129,
+      "loss": 0.6374,
+      "step": 864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012915,
+      "loss": 0.6809,
+      "step": 865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001293,
+      "loss": 0.7065,
+      "step": 866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012944999999999998,
+      "loss": 0.6583,
+      "step": 867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012959999999999998,
+      "loss": 0.8358,
+      "step": 868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012974999999999998,
+      "loss": 1.0078,
+      "step": 869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00012989999999999999,
+      "loss": 0.7145,
+      "step": 870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00013005,
+      "loss": 0.6655,
+      "step": 871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001302,
+      "loss": 0.6256,
+      "step": 872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00013035,
+      "loss": 0.6346,
+      "step": 873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001305,
+      "loss": 0.7322,
+      "step": 874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013064999999999998,
+      "loss": 1.2842,
+      "step": 875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013079999999999998,
+      "loss": 0.9512,
+      "step": 876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013094999999999998,
+      "loss": 0.7664,
+      "step": 877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001311,
+      "loss": 0.7621,
+      "step": 878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013125,
+      "loss": 0.6458,
+      "step": 879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001314,
+      "loss": 0.7962,
+      "step": 880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013155,
+      "loss": 0.7646,
+      "step": 881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013169999999999998,
+      "loss": 0.8169,
+      "step": 882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013184999999999998,
+      "loss": 0.7104,
+      "step": 883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 0.6716,
+      "step": 884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013215,
+      "loss": 0.6212,
+      "step": 885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001323,
+      "loss": 0.6458,
+      "step": 886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013245,
+      "loss": 0.8424,
+      "step": 887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001326,
+      "loss": 0.8954,
+      "step": 888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013275,
+      "loss": 0.9114,
+      "step": 889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013289999999999998,
+      "loss": 0.7036,
+      "step": 890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013304999999999998,
+      "loss": 0.7338,
+      "step": 891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013319999999999999,
+      "loss": 0.7343,
+      "step": 892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013335,
+      "loss": 0.7582,
+      "step": 893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001335,
+      "loss": 1.0635,
+      "step": 894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013365,
+      "loss": 0.7444,
+      "step": 895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001338,
+      "loss": 0.6428,
+      "step": 896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013395,
+      "loss": 0.6839,
+      "step": 897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013409999999999998,
+      "loss": 0.7789,
+      "step": 898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013424999999999998,
+      "loss": 0.7499,
+      "step": 899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001344,
+      "loss": 1.0531,
+      "step": 900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013455,
+      "loss": 0.7765,
+      "step": 901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001347,
+      "loss": 0.6512,
+      "step": 902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013485,
+      "loss": 0.6625,
+      "step": 903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000135,
+      "loss": 0.6059,
+      "step": 904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013514999999999998,
+      "loss": 0.6713,
+      "step": 905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013529999999999998,
+      "loss": 0.6967,
+      "step": 906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013544999999999999,
+      "loss": 0.8225,
+      "step": 907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001356,
+      "loss": 0.6438,
+      "step": 908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013575,
+      "loss": 0.6519,
+      "step": 909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001359,
+      "loss": 0.6664,
+      "step": 910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013605,
+      "loss": 0.6858,
+      "step": 911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001362,
+      "loss": 0.7617,
+      "step": 912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013634999999999998,
+      "loss": 1.1279,
+      "step": 913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013649999999999998,
+      "loss": 0.6765,
+      "step": 914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013665,
+      "loss": 0.772,
+      "step": 915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001368,
+      "loss": 0.7185,
+      "step": 916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013695,
+      "loss": 0.708,
+      "step": 917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001371,
+      "loss": 0.712,
+      "step": 918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013725,
+      "loss": 1.0943,
+      "step": 919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001374,
+      "loss": 0.6467,
+      "step": 920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013754999999999998,
+      "loss": 0.6327,
+      "step": 921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013769999999999999,
+      "loss": 0.7868,
+      "step": 922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013785,
+      "loss": 0.606,
+      "step": 923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000138,
+      "loss": 0.8671,
+      "step": 924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013815,
+      "loss": 0.9947,
+      "step": 925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001383,
+      "loss": 0.6948,
+      "step": 926
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013845,
+      "loss": 0.6984,
+      "step": 927
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001386,
+      "loss": 0.6391,
+      "step": 928
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013874999999999998,
+      "loss": 0.6413,
+      "step": 929
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001389,
+      "loss": 0.7014,
+      "step": 930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013905,
+      "loss": 0.7688,
+      "step": 931
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001392,
+      "loss": 0.8858,
+      "step": 932
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013935,
+      "loss": 0.6174,
+      "step": 933
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001395,
+      "loss": 0.5987,
+      "step": 934
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013965,
+      "loss": 0.6487,
+      "step": 935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013979999999999998,
+      "loss": 0.6604,
+      "step": 936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013995,
+      "loss": 0.8196,
+      "step": 937
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001401,
+      "loss": 0.9458,
+      "step": 938
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014025,
+      "loss": 0.6783,
+      "step": 939
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001404,
+      "loss": 0.7223,
+      "step": 940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014055,
+      "loss": 0.617,
+      "step": 941
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014069999999999998,
+      "loss": 0.7087,
+      "step": 942
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014084999999999998,
+      "loss": 0.7458,
+      "step": 943
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 1.0869,
+      "step": 944
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014115,
+      "loss": 0.6198,
+      "step": 945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001413,
+      "loss": 0.5765,
+      "step": 946
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014144999999999997,
+      "loss": 0.6038,
+      "step": 947
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014159999999999997,
+      "loss": 0.6967,
+      "step": 948
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014174999999999998,
+      "loss": 0.847,
+      "step": 949
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014189999999999998,
+      "loss": 1.2596,
+      "step": 950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014204999999999998,
+      "loss": 0.8075,
+      "step": 951
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001422,
+      "loss": 0.6227,
+      "step": 952
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014235,
+      "loss": 0.6795,
+      "step": 953
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001425,
+      "loss": 0.6252,
+      "step": 954
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014264999999999997,
+      "loss": 0.5868,
+      "step": 955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014279999999999997,
+      "loss": 0.7315,
+      "step": 956
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014294999999999998,
+      "loss": 0.875,
+      "step": 957
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014309999999999998,
+      "loss": 0.6182,
+      "step": 958
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014324999999999999,
+      "loss": 0.6437,
+      "step": 959
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001434,
+      "loss": 0.8841,
+      "step": 960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014355,
+      "loss": 0.7448,
+      "step": 961
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014369999999999997,
+      "loss": 0.6986,
+      "step": 962
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014384999999999997,
+      "loss": 0.8867,
+      "step": 963
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 0.5898,
+      "step": 964
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014414999999999998,
+      "loss": 0.6194,
+      "step": 965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014429999999999998,
+      "loss": 0.5922,
+      "step": 966
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014445,
+      "loss": 0.5534,
+      "step": 967
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001446,
+      "loss": 0.6801,
+      "step": 968
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014475,
+      "loss": 0.874,
+      "step": 969
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014489999999999997,
+      "loss": 0.6028,
+      "step": 970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014504999999999997,
+      "loss": 0.6365,
+      "step": 971
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014519999999999998,
+      "loss": 0.6322,
+      "step": 972
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014534999999999998,
+      "loss": 0.6751,
+      "step": 973
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014549999999999999,
+      "loss": 0.721,
+      "step": 974
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014565,
+      "loss": 0.7637,
+      "step": 975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001458,
+      "loss": 0.6825,
+      "step": 976
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014595,
+      "loss": 0.6271,
+      "step": 977
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014609999999999997,
+      "loss": 0.6345,
+      "step": 978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014624999999999998,
+      "loss": 0.741,
+      "step": 979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014639999999999998,
+      "loss": 0.6668,
+      "step": 980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014654999999999998,
+      "loss": 0.6861,
+      "step": 981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001467,
+      "loss": 0.9312,
+      "step": 982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014685,
+      "loss": 0.6118,
+      "step": 983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000147,
+      "loss": 0.6693,
+      "step": 984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014714999999999997,
+      "loss": 0.779,
+      "step": 985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014729999999999998,
+      "loss": 0.5938,
+      "step": 986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014744999999999998,
+      "loss": 0.7678,
+      "step": 987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014759999999999998,
+      "loss": 0.9229,
+      "step": 988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014774999999999999,
+      "loss": 0.6209,
+      "step": 989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001479,
+      "loss": 0.6224,
+      "step": 990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014805,
+      "loss": 0.8632,
+      "step": 991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001482,
+      "loss": 0.6261,
+      "step": 992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014834999999999997,
+      "loss": 0.7372,
+      "step": 993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014849999999999998,
+      "loss": 0.9951,
+      "step": 994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014864999999999998,
+      "loss": 0.6478,
+      "step": 995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014879999999999998,
+      "loss": 0.6383,
+      "step": 996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014895,
+      "loss": 0.6587,
+      "step": 997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001491,
+      "loss": 0.7199,
+      "step": 998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014925,
+      "loss": 0.6985,
+      "step": 999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001494,
+      "loss": 0.8429,
+      "step": 1000
+    },
+    {
+      "epoch": 0.38,
+      "eval_cer": 0.2022709093392122,
+      "eval_loss": 0.708672285079956,
+      "eval_runtime": 47.0095,
+      "eval_samples_per_second": 37.29,
+      "eval_steps_per_second": 4.68,
+      "eval_wer": 0.5028284382495362,
+      "step": 1000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014954999999999998,
+      "loss": 0.6724,
+      "step": 1001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014969999999999998,
+      "loss": 0.6005,
+      "step": 1002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014984999999999998,
+      "loss": 0.5893,
+      "step": 1003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015,
+      "loss": 0.5678,
+      "step": 1004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015014999999999996,
+      "loss": 0.6227,
+      "step": 1005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001503,
+      "loss": 0.7679,
+      "step": 1006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015044999999999997,
+      "loss": 0.6959,
+      "step": 1007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015059999999999997,
+      "loss": 0.5553,
+      "step": 1008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015074999999999998,
+      "loss": 0.5967,
+      "step": 1009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015089999999999998,
+      "loss": 0.6539,
+      "step": 1010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015104999999999996,
+      "loss": 0.6322,
+      "step": 1011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001512,
+      "loss": 0.6639,
+      "step": 1012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015134999999999997,
+      "loss": 0.8652,
+      "step": 1013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001515,
+      "loss": 0.6158,
+      "step": 1014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015164999999999997,
+      "loss": 0.6332,
+      "step": 1015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015179999999999998,
+      "loss": 0.6773,
+      "step": 1016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015194999999999998,
+      "loss": 0.5604,
+      "step": 1017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015209999999999998,
+      "loss": 0.5958,
+      "step": 1018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015224999999999996,
+      "loss": 0.9189,
+      "step": 1019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001524,
+      "loss": 0.5618,
+      "step": 1020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015254999999999997,
+      "loss": 0.599,
+      "step": 1021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001527,
+      "loss": 0.7571,
+      "step": 1022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015284999999999997,
+      "loss": 0.7081,
+      "step": 1023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015299999999999998,
+      "loss": 0.6801,
+      "step": 1024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015314999999999998,
+      "loss": 0.8832,
+      "step": 1025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015329999999999999,
+      "loss": 0.5583,
+      "step": 1026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015344999999999996,
+      "loss": 0.732,
+      "step": 1027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001536,
+      "loss": 0.6083,
+      "step": 1028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015374999999999997,
+      "loss": 0.5765,
+      "step": 1029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001539,
+      "loss": 0.5758,
+      "step": 1030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015404999999999998,
+      "loss": 0.7559,
+      "step": 1031
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015419999999999998,
+      "loss": 0.7921,
+      "step": 1032
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015434999999999998,
+      "loss": 0.5844,
+      "step": 1033
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001545,
+      "loss": 0.6965,
+      "step": 1034
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015464999999999996,
+      "loss": 0.5794,
+      "step": 1035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001548,
+      "loss": 0.6973,
+      "step": 1036
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015494999999999997,
+      "loss": 0.7105,
+      "step": 1037
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001551,
+      "loss": 0.8945,
+      "step": 1038
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015524999999999998,
+      "loss": 0.6082,
+      "step": 1039
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015539999999999998,
+      "loss": 0.5414,
+      "step": 1040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015554999999999999,
+      "loss": 0.5753,
+      "step": 1041
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001557,
+      "loss": 0.5301,
+      "step": 1042
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015584999999999997,
+      "loss": 0.5734,
+      "step": 1043
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000156,
+      "loss": 0.9239,
+      "step": 1044
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015614999999999997,
+      "loss": 0.5208,
+      "step": 1045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001563,
+      "loss": 0.6177,
+      "step": 1046
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015644999999999998,
+      "loss": 0.7747,
+      "step": 1047
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 0.6003,
+      "step": 1048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015675,
+      "loss": 0.811,
+      "step": 1049
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001569,
+      "loss": 0.8188,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015704999999999997,
+      "loss": 0.5969,
+      "step": 1051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001572,
+      "loss": 0.5535,
+      "step": 1052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015734999999999998,
+      "loss": 0.537,
+      "step": 1053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015749999999999998,
+      "loss": 0.6247,
+      "step": 1054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015764999999999998,
+      "loss": 0.6055,
+      "step": 1055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001578,
+      "loss": 0.7304,
+      "step": 1056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00015794999999999996,
+      "loss": 0.8844,
+      "step": 1057
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001581,
+      "loss": 0.6978,
+      "step": 1058
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00015824999999999997,
+      "loss": 0.5855,
+      "step": 1059
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001584,
+      "loss": 0.566,
+      "step": 1060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00015854999999999998,
+      "loss": 0.5379,
+      "step": 1061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00015869999999999998,
+      "loss": 0.7971,
+      "step": 1062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00015884999999999999,
+      "loss": 0.996,
+      "step": 1063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000159,
+      "loss": 0.5896,
+      "step": 1064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00015914999999999997,
+      "loss": 0.5452,
+      "step": 1065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001593,
+      "loss": 0.8378,
+      "step": 1066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00015944999999999997,
+      "loss": 0.5666,
+      "step": 1067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001596,
+      "loss": 0.5953,
+      "step": 1068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00015974999999999998,
+      "loss": 0.9395,
+      "step": 1069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00015989999999999998,
+      "loss": 0.5724,
+      "step": 1070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016005,
+      "loss": 0.6362,
+      "step": 1071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001602,
+      "loss": 0.7034,
+      "step": 1072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016034999999999997,
+      "loss": 0.6764,
+      "step": 1073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001605,
+      "loss": 0.6078,
+      "step": 1074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016064999999999997,
+      "loss": 0.9673,
+      "step": 1075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001608,
+      "loss": 0.5675,
+      "step": 1076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016094999999999998,
+      "loss": 0.7448,
+      "step": 1077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016109999999999999,
+      "loss": 0.5362,
+      "step": 1078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016125,
+      "loss": 0.5568,
+      "step": 1079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001614,
+      "loss": 0.672,
+      "step": 1080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00016154999999999997,
+      "loss": 0.8592,
+      "step": 1081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001617,
+      "loss": 0.9695,
+      "step": 1082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016184999999999998,
+      "loss": 0.8051,
+      "step": 1083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000162,
+      "loss": 0.6126,
+      "step": 1084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016214999999999998,
+      "loss": 0.4866,
+      "step": 1085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001623,
+      "loss": 0.627,
+      "step": 1086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016245,
+      "loss": 0.6422,
+      "step": 1087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001626,
+      "loss": 0.8685,
+      "step": 1088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016274999999999997,
+      "loss": 0.6309,
+      "step": 1089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001629,
+      "loss": 0.8615,
+      "step": 1090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016304999999999998,
+      "loss": 0.5984,
+      "step": 1091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001632,
+      "loss": 0.563,
+      "step": 1092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016334999999999999,
+      "loss": 0.735,
+      "step": 1093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001635,
+      "loss": 0.7278,
+      "step": 1094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016365,
+      "loss": 0.5556,
+      "step": 1095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001638,
+      "loss": 0.5695,
+      "step": 1096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016394999999999997,
+      "loss": 0.7134,
+      "step": 1097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001641,
+      "loss": 0.5432,
+      "step": 1098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016424999999999998,
+      "loss": 0.5699,
+      "step": 1099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001644,
+      "loss": 1.2932,
+      "step": 1100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016455,
+      "loss": 0.6399,
+      "step": 1101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001647,
+      "loss": 0.6078,
+      "step": 1102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016485,
+      "loss": 0.6822,
+      "step": 1103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000165,
+      "loss": 0.5511,
+      "step": 1104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016514999999999998,
+      "loss": 0.5833,
+      "step": 1105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001653,
+      "loss": 0.7879,
+      "step": 1106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00016544999999999998,
+      "loss": 0.8734,
+      "step": 1107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001656,
+      "loss": 0.5293,
+      "step": 1108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016575,
+      "loss": 0.4606,
+      "step": 1109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001659,
+      "loss": 0.5157,
+      "step": 1110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016604999999999997,
+      "loss": 0.5646,
+      "step": 1111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001662,
+      "loss": 0.6473,
+      "step": 1112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016634999999999998,
+      "loss": 0.6556,
+      "step": 1113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001665,
+      "loss": 0.5143,
+      "step": 1114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016664999999999998,
+      "loss": 0.6298,
+      "step": 1115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001668,
+      "loss": 0.5343,
+      "step": 1116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016695,
+      "loss": 0.6701,
+      "step": 1117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001671,
+      "loss": 0.6127,
+      "step": 1118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016724999999999997,
+      "loss": 1.0177,
+      "step": 1119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001674,
+      "loss": 0.7399,
+      "step": 1120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016754999999999998,
+      "loss": 0.6348,
+      "step": 1121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001677,
+      "loss": 0.6461,
+      "step": 1122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016785,
+      "loss": 0.4557,
+      "step": 1123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000168,
+      "loss": 0.7338,
+      "step": 1124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016815,
+      "loss": 0.8664,
+      "step": 1125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001683,
+      "loss": 0.5353,
+      "step": 1126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016844999999999997,
+      "loss": 0.603,
+      "step": 1127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001686,
+      "loss": 0.634,
+      "step": 1128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016874999999999998,
+      "loss": 0.7307,
+      "step": 1129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016889999999999996,
+      "loss": 0.58,
+      "step": 1130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016905,
+      "loss": 0.735,
+      "step": 1131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016919999999999997,
+      "loss": 0.7578,
+      "step": 1132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016935,
+      "loss": 0.6327,
+      "step": 1133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00016949999999999997,
+      "loss": 0.5269,
+      "step": 1134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016964999999999998,
+      "loss": 0.5235,
+      "step": 1135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016979999999999998,
+      "loss": 0.5628,
+      "step": 1136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00016994999999999998,
+      "loss": 0.58,
+      "step": 1137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017009999999999996,
+      "loss": 0.7507,
+      "step": 1138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017025,
+      "loss": 0.5782,
+      "step": 1139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017039999999999997,
+      "loss": 0.57,
+      "step": 1140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017055,
+      "loss": 0.4967,
+      "step": 1141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017069999999999998,
+      "loss": 0.5967,
+      "step": 1142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017084999999999998,
+      "loss": 0.6624,
+      "step": 1143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017099999999999998,
+      "loss": 0.754,
+      "step": 1144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017114999999999999,
+      "loss": 0.5388,
+      "step": 1145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017129999999999996,
+      "loss": 0.4924,
+      "step": 1146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017145,
+      "loss": 0.6261,
+      "step": 1147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017159999999999997,
+      "loss": 0.5818,
+      "step": 1148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017175,
+      "loss": 0.5524,
+      "step": 1149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017189999999999998,
+      "loss": 0.7286,
+      "step": 1150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017204999999999998,
+      "loss": 0.5225,
+      "step": 1151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017219999999999998,
+      "loss": 0.6506,
+      "step": 1152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017235,
+      "loss": 0.5215,
+      "step": 1153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017249999999999996,
+      "loss": 0.5482,
+      "step": 1154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017265,
+      "loss": 0.4838,
+      "step": 1155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017279999999999997,
+      "loss": 0.6632,
+      "step": 1156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017294999999999998,
+      "loss": 0.6384,
+      "step": 1157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017309999999999998,
+      "loss": 0.7183,
+      "step": 1158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017324999999999998,
+      "loss": 0.5214,
+      "step": 1159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017339999999999996,
+      "loss": 0.5375,
+      "step": 1160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00017355,
+      "loss": 0.6201,
+      "step": 1161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017369999999999997,
+      "loss": 0.6281,
+      "step": 1162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017385,
+      "loss": 0.7992,
+      "step": 1163
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 0.5814,
+      "step": 1164
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017414999999999998,
+      "loss": 0.4603,
+      "step": 1165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017429999999999998,
+      "loss": 0.5079,
+      "step": 1166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017444999999999998,
+      "loss": 0.5527,
+      "step": 1167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017459999999999996,
+      "loss": 0.5575,
+      "step": 1168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017475,
+      "loss": 0.7477,
+      "step": 1169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017489999999999997,
+      "loss": 0.5175,
+      "step": 1170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017505,
+      "loss": 0.5974,
+      "step": 1171
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017519999999999998,
+      "loss": 0.4935,
+      "step": 1172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017534999999999998,
+      "loss": 0.5039,
+      "step": 1173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017549999999999998,
+      "loss": 0.4632,
+      "step": 1174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017565,
+      "loss": 0.6322,
+      "step": 1175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017579999999999996,
+      "loss": 0.4478,
+      "step": 1176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017595,
+      "loss": 0.5728,
+      "step": 1177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017609999999999997,
+      "loss": 0.4308,
+      "step": 1178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017625,
+      "loss": 0.487,
+      "step": 1179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017639999999999998,
+      "loss": 0.6819,
+      "step": 1180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017654999999999998,
+      "loss": 0.6619,
+      "step": 1181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017669999999999999,
+      "loss": 0.6362,
+      "step": 1182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017685,
+      "loss": 0.4678,
+      "step": 1183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 0.4754,
+      "step": 1184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017715,
+      "loss": 0.5046,
+      "step": 1185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017729999999999997,
+      "loss": 0.5469,
+      "step": 1186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00017745,
+      "loss": 0.5373,
+      "step": 1187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017759999999999998,
+      "loss": 0.7442,
+      "step": 1188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017774999999999998,
+      "loss": 0.6132,
+      "step": 1189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001779,
+      "loss": 0.5317,
+      "step": 1190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017805,
+      "loss": 0.4178,
+      "step": 1191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017819999999999997,
+      "loss": 0.5321,
+      "step": 1192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017835,
+      "loss": 0.5189,
+      "step": 1193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017849999999999997,
+      "loss": 0.8278,
+      "step": 1194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017865,
+      "loss": 0.5572,
+      "step": 1195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017879999999999998,
+      "loss": 0.4442,
+      "step": 1196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017894999999999999,
+      "loss": 0.6327,
+      "step": 1197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001791,
+      "loss": 0.4592,
+      "step": 1198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017925,
+      "loss": 0.5448,
+      "step": 1199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017939999999999997,
+      "loss": 0.7,
+      "step": 1200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017955,
+      "loss": 0.6672,
+      "step": 1201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017969999999999998,
+      "loss": 0.5314,
+      "step": 1202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017984999999999998,
+      "loss": 0.4826,
+      "step": 1203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 0.5077,
+      "step": 1204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018015,
+      "loss": 0.5544,
+      "step": 1205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018029999999999996,
+      "loss": 0.6454,
+      "step": 1206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018045,
+      "loss": 0.8642,
+      "step": 1207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018059999999999997,
+      "loss": 0.5262,
+      "step": 1208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018075,
+      "loss": 0.5776,
+      "step": 1209
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018089999999999998,
+      "loss": 0.4969,
+      "step": 1210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018104999999999998,
+      "loss": 0.5057,
+      "step": 1211
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018119999999999999,
+      "loss": 0.5133,
+      "step": 1212
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018135,
+      "loss": 1.1167,
+      "step": 1213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018149999999999997,
+      "loss": 0.5013,
+      "step": 1214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018165,
+      "loss": 0.4735,
+      "step": 1215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018179999999999997,
+      "loss": 0.5573,
+      "step": 1216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018195,
+      "loss": 0.4749,
+      "step": 1217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018209999999999998,
+      "loss": 0.6465,
+      "step": 1218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018224999999999998,
+      "loss": 0.6497,
+      "step": 1219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001824,
+      "loss": 0.4953,
+      "step": 1220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018255,
+      "loss": 0.4767,
+      "step": 1221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018269999999999997,
+      "loss": 0.5054,
+      "step": 1222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018285,
+      "loss": 0.4891,
+      "step": 1223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018299999999999998,
+      "loss": 0.5884,
+      "step": 1224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018315,
+      "loss": 0.7378,
+      "step": 1225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018329999999999998,
+      "loss": 0.5677,
+      "step": 1226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018345,
+      "loss": 0.4364,
+      "step": 1227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001836,
+      "loss": 0.4586,
+      "step": 1228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018375,
+      "loss": 0.542,
+      "step": 1229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018389999999999997,
+      "loss": 0.4672,
+      "step": 1230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018405,
+      "loss": 0.7332,
+      "step": 1231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018419999999999998,
+      "loss": 1.1025,
+      "step": 1232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018435,
+      "loss": 0.6345,
+      "step": 1233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018449999999999999,
+      "loss": 0.4316,
+      "step": 1234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018465,
+      "loss": 0.4932,
+      "step": 1235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001848,
+      "loss": 0.5166,
+      "step": 1236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018495,
+      "loss": 0.5982,
+      "step": 1237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018509999999999997,
+      "loss": 0.7333,
+      "step": 1238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018525,
+      "loss": 0.5411,
+      "step": 1239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018539999999999998,
+      "loss": 0.6048,
+      "step": 1240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018555,
+      "loss": 0.4734,
+      "step": 1241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001857,
+      "loss": 0.4915,
+      "step": 1242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018585,
+      "loss": 0.5279,
+      "step": 1243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000186,
+      "loss": 0.9114,
+      "step": 1244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018615,
+      "loss": 0.6011,
+      "step": 1245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018629999999999997,
+      "loss": 0.6099,
+      "step": 1246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018645,
+      "loss": 0.4977,
+      "step": 1247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018659999999999998,
+      "loss": 0.5411,
+      "step": 1248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018675,
+      "loss": 0.4817,
+      "step": 1249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001869,
+      "loss": 0.7342,
+      "step": 1250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018705,
+      "loss": 0.9459,
+      "step": 1251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001872,
+      "loss": 0.5107,
+      "step": 1252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018735,
+      "loss": 0.5678,
+      "step": 1253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018749999999999998,
+      "loss": 0.4419,
+      "step": 1254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018764999999999998,
+      "loss": 0.4539,
+      "step": 1255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018779999999999998,
+      "loss": 0.6498,
+      "step": 1256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018794999999999996,
+      "loss": 0.633,
+      "step": 1257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001881,
+      "loss": 0.6435,
+      "step": 1258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018824999999999997,
+      "loss": 0.4895,
+      "step": 1259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018839999999999997,
+      "loss": 0.7503,
+      "step": 1260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018854999999999998,
+      "loss": 0.5609,
+      "step": 1261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018869999999999998,
+      "loss": 0.5921,
+      "step": 1262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018884999999999996,
+      "loss": 0.6619,
+      "step": 1263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 0.5688,
+      "step": 1264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018914999999999996,
+      "loss": 0.4343,
+      "step": 1265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001893,
+      "loss": 0.5863,
+      "step": 1266
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018944999999999997,
+      "loss": 0.472,
+      "step": 1267
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018959999999999997,
+      "loss": 0.5776,
+      "step": 1268
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018974999999999998,
+      "loss": 0.7596,
+      "step": 1269
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018989999999999998,
+      "loss": 0.573,
+      "step": 1270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019004999999999996,
+      "loss": 0.4689,
+      "step": 1271
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001902,
+      "loss": 0.5322,
+      "step": 1272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019034999999999996,
+      "loss": 0.4998,
+      "step": 1273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001905,
+      "loss": 0.6281,
+      "step": 1274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019064999999999997,
+      "loss": 0.8379,
+      "step": 1275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019079999999999998,
+      "loss": 0.5222,
+      "step": 1276
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019094999999999998,
+      "loss": 0.4885,
+      "step": 1277
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019109999999999998,
+      "loss": 0.6439,
+      "step": 1278
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019124999999999996,
+      "loss": 0.504,
+      "step": 1279
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001914,
+      "loss": 0.6116,
+      "step": 1280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019154999999999997,
+      "loss": 0.7225,
+      "step": 1281
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001917,
+      "loss": 0.6816,
+      "step": 1282
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019184999999999997,
+      "loss": 0.4458,
+      "step": 1283
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 0.6328,
+      "step": 1284
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019214999999999998,
+      "loss": 0.4995,
+      "step": 1285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019229999999999999,
+      "loss": 0.4286,
+      "step": 1286
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019244999999999996,
+      "loss": 0.5023,
+      "step": 1287
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001926,
+      "loss": 0.7743,
+      "step": 1288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019274999999999997,
+      "loss": 0.5167,
+      "step": 1289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001929,
+      "loss": 0.532,
+      "step": 1290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019304999999999998,
+      "loss": 0.406,
+      "step": 1291
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019319999999999998,
+      "loss": 0.5039,
+      "step": 1292
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019334999999999998,
+      "loss": 0.4507,
+      "step": 1293
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001935,
+      "loss": 0.7308,
+      "step": 1294
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019364999999999996,
+      "loss": 0.4562,
+      "step": 1295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001938,
+      "loss": 0.4105,
+      "step": 1296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019394999999999997,
+      "loss": 0.5752,
+      "step": 1297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001941,
+      "loss": 0.4516,
+      "step": 1298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019424999999999998,
+      "loss": 0.4563,
+      "step": 1299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019439999999999998,
+      "loss": 0.809,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019454999999999999,
+      "loss": 0.4113,
+      "step": 1301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001947,
+      "loss": 0.5158,
+      "step": 1302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019484999999999997,
+      "loss": 0.4731,
+      "step": 1303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000195,
+      "loss": 0.4635,
+      "step": 1304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019514999999999997,
+      "loss": 0.4777,
+      "step": 1305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019529999999999998,
+      "loss": 0.6833,
+      "step": 1306
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019544999999999998,
+      "loss": 0.6283,
+      "step": 1307
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019559999999999998,
+      "loss": 0.4961,
+      "step": 1308
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019574999999999996,
+      "loss": 0.5197,
+      "step": 1309
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001959,
+      "loss": 0.4153,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019604999999999997,
+      "loss": 0.4619,
+      "step": 1311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001962,
+      "loss": 0.5015,
+      "step": 1312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019634999999999998,
+      "loss": 0.7153,
+      "step": 1313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019649999999999998,
+      "loss": 0.5959,
+      "step": 1314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019664999999999998,
+      "loss": 0.4058,
+      "step": 1315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019679999999999999,
+      "loss": 0.458,
+      "step": 1316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019694999999999996,
+      "loss": 0.3972,
+      "step": 1317
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001971,
+      "loss": 0.4421,
+      "step": 1318
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019724999999999997,
+      "loss": 0.7187,
+      "step": 1319
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001974,
+      "loss": 0.4424,
+      "step": 1320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019754999999999998,
+      "loss": 0.4235,
+      "step": 1321
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019769999999999998,
+      "loss": 0.4104,
+      "step": 1322
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019784999999999998,
+      "loss": 0.3624,
+      "step": 1323
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000198,
+      "loss": 0.5169,
+      "step": 1324
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019814999999999996,
+      "loss": 0.8557,
+      "step": 1325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001983,
+      "loss": 0.4314,
+      "step": 1326
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019844999999999997,
+      "loss": 0.4623,
+      "step": 1327
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001986,
+      "loss": 0.4628,
+      "step": 1328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019874999999999998,
+      "loss": 0.4799,
+      "step": 1329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019889999999999998,
+      "loss": 0.4506,
+      "step": 1330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019905,
+      "loss": 0.595,
+      "step": 1331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001992,
+      "loss": 0.6821,
+      "step": 1332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019934999999999997,
+      "loss": 0.5324,
+      "step": 1333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001995,
+      "loss": 0.4968,
+      "step": 1334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019964999999999997,
+      "loss": 0.4438,
+      "step": 1335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001998,
+      "loss": 0.4806,
+      "step": 1336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019994999999999998,
+      "loss": 0.4444,
+      "step": 1337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020009999999999998,
+      "loss": 0.7046,
+      "step": 1338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020025,
+      "loss": 0.4305,
+      "step": 1339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002004,
+      "loss": 0.3481,
+      "step": 1340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020054999999999997,
+      "loss": 0.4475,
+      "step": 1341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002007,
+      "loss": 0.4752,
+      "step": 1342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00020084999999999998,
+      "loss": 0.5168,
+      "step": 1343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000201,
+      "loss": 0.7344,
+      "step": 1344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020114999999999998,
+      "loss": 0.4423,
+      "step": 1345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002013,
+      "loss": 0.4975,
+      "step": 1346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020145,
+      "loss": 0.4217,
+      "step": 1347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002016,
+      "loss": 0.4529,
+      "step": 1348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020174999999999997,
+      "loss": 0.5296,
+      "step": 1349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002019,
+      "loss": 0.7826,
+      "step": 1350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020204999999999998,
+      "loss": 0.4126,
+      "step": 1351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002022,
+      "loss": 0.36,
+      "step": 1352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020234999999999999,
+      "loss": 0.3884,
+      "step": 1353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002025,
+      "loss": 0.3903,
+      "step": 1354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020264999999999997,
+      "loss": 0.6517,
+      "step": 1355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002028,
+      "loss": 0.5192,
+      "step": 1356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020294999999999997,
+      "loss": 0.9808,
+      "step": 1357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002031,
+      "loss": 0.4537,
+      "step": 1358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020324999999999998,
+      "loss": 0.4427,
+      "step": 1359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020339999999999998,
+      "loss": 0.4514,
+      "step": 1360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020355,
+      "loss": 0.3743,
+      "step": 1361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002037,
+      "loss": 0.6618,
+      "step": 1362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020384999999999997,
+      "loss": 0.7594,
+      "step": 1363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000204,
+      "loss": 0.3746,
+      "step": 1364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020414999999999997,
+      "loss": 0.3718,
+      "step": 1365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002043,
+      "loss": 0.3641,
+      "step": 1366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020444999999999998,
+      "loss": 0.4923,
+      "step": 1367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020459999999999999,
+      "loss": 0.5157,
+      "step": 1368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00020475,
+      "loss": 0.7151,
+      "step": 1369
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002049,
+      "loss": 0.4354,
+      "step": 1370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020504999999999997,
+      "loss": 0.3597,
+      "step": 1371
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002052,
+      "loss": 0.4548,
+      "step": 1372
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020534999999999998,
+      "loss": 0.3263,
+      "step": 1373
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002055,
+      "loss": 0.4734,
+      "step": 1374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020564999999999998,
+      "loss": 0.8631,
+      "step": 1375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002058,
+      "loss": 0.5101,
+      "step": 1376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020595,
+      "loss": 0.4569,
+      "step": 1377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002061,
+      "loss": 0.4482,
+      "step": 1378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020624999999999997,
+      "loss": 0.3863,
+      "step": 1379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020639999999999998,
+      "loss": 0.3461,
+      "step": 1380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020654999999999998,
+      "loss": 0.4857,
+      "step": 1381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020669999999999996,
+      "loss": 0.6993,
+      "step": 1382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020684999999999999,
+      "loss": 0.3722,
+      "step": 1383
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 0.5422,
+      "step": 1384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020715,
+      "loss": 0.5472,
+      "step": 1385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020729999999999997,
+      "loss": 0.4422,
+      "step": 1386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020744999999999997,
+      "loss": 0.4367,
+      "step": 1387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020759999999999998,
+      "loss": 0.5962,
+      "step": 1388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020774999999999998,
+      "loss": 0.5099,
+      "step": 1389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020789999999999996,
+      "loss": 0.4237,
+      "step": 1390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020805,
+      "loss": 0.3642,
+      "step": 1391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020819999999999996,
+      "loss": 0.3761,
+      "step": 1392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020835,
+      "loss": 0.4183,
+      "step": 1393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020849999999999997,
+      "loss": 0.6532,
+      "step": 1394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00020864999999999998,
+      "loss": 0.3615,
+      "step": 1395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00020879999999999998,
+      "loss": 0.6322,
+      "step": 1396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00020894999999999998,
+      "loss": 0.3768,
+      "step": 1397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00020909999999999996,
+      "loss": 0.4072,
+      "step": 1398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00020925,
+      "loss": 0.3914,
+      "step": 1399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00020939999999999997,
+      "loss": 0.5885,
+      "step": 1400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00020955,
+      "loss": 0.367,
+      "step": 1401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00020969999999999997,
+      "loss": 0.4043,
+      "step": 1402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00020984999999999998,
+      "loss": 0.5285,
+      "step": 1403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 0.3702,
+      "step": 1404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021014999999999999,
+      "loss": 0.4666,
+      "step": 1405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021029999999999996,
+      "loss": 0.5173,
+      "step": 1406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021045,
+      "loss": 0.6291,
+      "step": 1407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021059999999999997,
+      "loss": 0.452,
+      "step": 1408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021074999999999997,
+      "loss": 0.3508,
+      "step": 1409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021089999999999998,
+      "loss": 0.359,
+      "step": 1410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021104999999999998,
+      "loss": 0.3796,
+      "step": 1411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021119999999999996,
+      "loss": 0.419,
+      "step": 1412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021135,
+      "loss": 0.8116,
+      "step": 1413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021149999999999996,
+      "loss": 0.7631,
+      "step": 1414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021165,
+      "loss": 0.4144,
+      "step": 1415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021179999999999997,
+      "loss": 0.4448,
+      "step": 1416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021194999999999997,
+      "loss": 0.3834,
+      "step": 1417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021209999999999998,
+      "loss": 0.3176,
+      "step": 1418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021224999999999998,
+      "loss": 0.5204,
+      "step": 1419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021239999999999996,
+      "loss": 0.3724,
+      "step": 1420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00021255,
+      "loss": 0.4156,
+      "step": 1421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021269999999999997,
+      "loss": 0.3584,
+      "step": 1422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021285,
+      "loss": 0.4718,
+      "step": 1423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 0.3444,
+      "step": 1424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021314999999999998,
+      "loss": 0.6328,
+      "step": 1425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021329999999999998,
+      "loss": 0.4563,
+      "step": 1426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021344999999999998,
+      "loss": 0.6701,
+      "step": 1427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021359999999999996,
+      "loss": 0.3333,
+      "step": 1428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021375,
+      "loss": 0.3341,
+      "step": 1429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021389999999999997,
+      "loss": 0.4795,
+      "step": 1430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021405,
+      "loss": 0.5713,
+      "step": 1431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021419999999999998,
+      "loss": 0.6026,
+      "step": 1432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021434999999999998,
+      "loss": 0.4604,
+      "step": 1433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021449999999999998,
+      "loss": 0.4848,
+      "step": 1434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021464999999999999,
+      "loss": 0.4333,
+      "step": 1435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021479999999999996,
+      "loss": 0.4254,
+      "step": 1436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021495,
+      "loss": 0.4501,
+      "step": 1437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021509999999999997,
+      "loss": 0.6539,
+      "step": 1438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021525,
+      "loss": 0.3718,
+      "step": 1439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021539999999999998,
+      "loss": 0.3605,
+      "step": 1440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021554999999999998,
+      "loss": 0.4973,
+      "step": 1441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021569999999999998,
+      "loss": 0.3709,
+      "step": 1442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021585,
+      "loss": 0.3343,
+      "step": 1443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 0.6312,
+      "step": 1444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021615,
+      "loss": 0.391,
+      "step": 1445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021629999999999997,
+      "loss": 0.4484,
+      "step": 1446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00021645,
+      "loss": 0.403,
+      "step": 1447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021659999999999998,
+      "loss": 0.4355,
+      "step": 1448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021674999999999998,
+      "loss": 0.4752,
+      "step": 1449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002169,
+      "loss": 0.8421,
+      "step": 1450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021705,
+      "loss": 0.4561,
+      "step": 1451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021719999999999997,
+      "loss": 0.4786,
+      "step": 1452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021735,
+      "loss": 0.4241,
+      "step": 1453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021749999999999997,
+      "loss": 0.333,
+      "step": 1454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021764999999999998,
+      "loss": 0.3477,
+      "step": 1455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021779999999999998,
+      "loss": 0.4093,
+      "step": 1456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021794999999999999,
+      "loss": 0.4919,
+      "step": 1457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021809999999999996,
+      "loss": 0.309,
+      "step": 1458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021825,
+      "loss": 0.4266,
+      "step": 1459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021839999999999997,
+      "loss": 0.3167,
+      "step": 1460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021855,
+      "loss": 0.357,
+      "step": 1461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021869999999999998,
+      "loss": 0.546,
+      "step": 1462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021884999999999998,
+      "loss": 0.6648,
+      "step": 1463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 0.4039,
+      "step": 1464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021915,
+      "loss": 0.3798,
+      "step": 1465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021929999999999996,
+      "loss": 0.3764,
+      "step": 1466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021945,
+      "loss": 0.4826,
+      "step": 1467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021959999999999997,
+      "loss": 0.4047,
+      "step": 1468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021975,
+      "loss": 0.6253,
+      "step": 1469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00021989999999999998,
+      "loss": 0.4207,
+      "step": 1470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00022004999999999998,
+      "loss": 0.4762,
+      "step": 1471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00022019999999999999,
+      "loss": 0.3912,
+      "step": 1472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00022035,
+      "loss": 0.3585,
+      "step": 1473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00022049999999999997,
+      "loss": 0.5611,
+      "step": 1474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022065,
+      "loss": 0.5776,
+      "step": 1475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022079999999999997,
+      "loss": 0.4299,
+      "step": 1476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022095,
+      "loss": 0.36,
+      "step": 1477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022109999999999998,
+      "loss": 0.6153,
+      "step": 1478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022124999999999998,
+      "loss": 0.3393,
+      "step": 1479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002214,
+      "loss": 0.3227,
+      "step": 1480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022155,
+      "loss": 0.5884,
+      "step": 1481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022169999999999997,
+      "loss": 0.5589,
+      "step": 1482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022185,
+      "loss": 0.3249,
+      "step": 1483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 0.3929,
+      "step": 1484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022215,
+      "loss": 0.4031,
+      "step": 1485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022229999999999998,
+      "loss": 0.643,
+      "step": 1486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022244999999999999,
+      "loss": 0.4627,
+      "step": 1487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002226,
+      "loss": 0.5382,
+      "step": 1488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022275,
+      "loss": 0.9632,
+      "step": 1489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022289999999999997,
+      "loss": 0.4138,
+      "step": 1490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022305,
+      "loss": 0.583,
+      "step": 1491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022319999999999998,
+      "loss": 0.4842,
+      "step": 1492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022335,
+      "loss": 0.4082,
+      "step": 1493
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022349999999999998,
+      "loss": 0.773,
+      "step": 1494
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022365,
+      "loss": 0.3573,
+      "step": 1495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002238,
+      "loss": 0.3221,
+      "step": 1496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022395,
+      "loss": 0.4256,
+      "step": 1497
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022409999999999997,
+      "loss": 0.4094,
+      "step": 1498
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022425,
+      "loss": 0.4636,
+      "step": 1499
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00022439999999999998,
+      "loss": 0.7377,
+      "step": 1500
+    },
+    {
+      "epoch": 0.57,
+      "eval_cer": 0.13394073004662777,
+      "eval_loss": 0.4899697005748749,
+      "eval_runtime": 47.7706,
+      "eval_samples_per_second": 36.696,
+      "eval_steps_per_second": 4.605,
+      "eval_wer": 0.27784314612843375,
+      "step": 1500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022455,
+      "loss": 0.3593,
+      "step": 1501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002247,
+      "loss": 0.4253,
+      "step": 1502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022485,
+      "loss": 0.5078,
+      "step": 1503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000225,
+      "loss": 0.3864,
+      "step": 1504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022514999999999997,
+      "loss": 0.4961,
+      "step": 1505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022529999999999997,
+      "loss": 0.4443,
+      "step": 1506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022544999999999995,
+      "loss": 0.5614,
+      "step": 1507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022559999999999998,
+      "loss": 0.3015,
+      "step": 1508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022574999999999996,
+      "loss": 0.3669,
+      "step": 1509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002259,
+      "loss": 0.3568,
+      "step": 1510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022604999999999997,
+      "loss": 0.3665,
+      "step": 1511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022619999999999997,
+      "loss": 0.6162,
+      "step": 1512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022634999999999997,
+      "loss": 0.5802,
+      "step": 1513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022649999999999998,
+      "loss": 0.3265,
+      "step": 1514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022664999999999995,
+      "loss": 0.5248,
+      "step": 1515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022679999999999998,
+      "loss": 0.3713,
+      "step": 1516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022694999999999996,
+      "loss": 0.4262,
+      "step": 1517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002271,
+      "loss": 0.4028,
+      "step": 1518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022724999999999997,
+      "loss": 0.6943,
+      "step": 1519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022739999999999997,
+      "loss": 0.3163,
+      "step": 1520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022754999999999997,
+      "loss": 0.3807,
+      "step": 1521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022769999999999998,
+      "loss": 0.3529,
+      "step": 1522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022784999999999995,
+      "loss": 0.4411,
+      "step": 1523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 0.371,
+      "step": 1524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00022814999999999996,
+      "loss": 0.6276,
+      "step": 1525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002283,
+      "loss": 0.5598,
+      "step": 1526
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022844999999999997,
+      "loss": 0.3943,
+      "step": 1527
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022859999999999997,
+      "loss": 0.3638,
+      "step": 1528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022874999999999998,
+      "loss": 0.3422,
+      "step": 1529
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022889999999999998,
+      "loss": 0.3876,
+      "step": 1530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022904999999999996,
+      "loss": 0.4418,
+      "step": 1531
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002292,
+      "loss": 0.564,
+      "step": 1532
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022934999999999996,
+      "loss": 0.3144,
+      "step": 1533
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002295,
+      "loss": 0.3453,
+      "step": 1534
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022964999999999997,
+      "loss": 0.4507,
+      "step": 1535
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022979999999999997,
+      "loss": 0.582,
+      "step": 1536
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00022994999999999998,
+      "loss": 0.5392,
+      "step": 1537
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00023009999999999998,
+      "loss": 0.537,
+      "step": 1538
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00023024999999999996,
+      "loss": 0.3536,
+      "step": 1539
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002304,
+      "loss": 0.3292,
+      "step": 1540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00023054999999999997,
+      "loss": 0.3574,
+      "step": 1541
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002307,
+      "loss": 0.3351,
+      "step": 1542
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00023084999999999997,
+      "loss": 0.4989,
+      "step": 1543
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00023099999999999998,
+      "loss": 0.6478,
+      "step": 1544
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00023114999999999998,
+      "loss": 0.581,
+      "step": 1545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00023129999999999998,
+      "loss": 0.3783,
+      "step": 1546
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00023144999999999996,
+      "loss": 0.3428,
+      "step": 1547
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002316,
+      "loss": 0.5025,
+      "step": 1548
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00023174999999999997,
+      "loss": 0.351,
+      "step": 1549
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002319,
+      "loss": 0.6288,
+      "step": 1550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00023204999999999998,
+      "loss": 0.336,
+      "step": 1551
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00023219999999999998,
+      "loss": 0.3513,
+      "step": 1552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023234999999999998,
+      "loss": 0.285,
+      "step": 1553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023249999999999999,
+      "loss": 0.355,
+      "step": 1554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023264999999999996,
+      "loss": 0.553,
+      "step": 1555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002328,
+      "loss": 0.4046,
+      "step": 1556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023294999999999997,
+      "loss": 0.5468,
+      "step": 1557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023309999999999997,
+      "loss": 0.3777,
+      "step": 1558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023324999999999998,
+      "loss": 0.4199,
+      "step": 1559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023339999999999998,
+      "loss": 0.3687,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023354999999999996,
+      "loss": 0.3575,
+      "step": 1561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002337,
+      "loss": 0.4144,
+      "step": 1562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023384999999999997,
+      "loss": 0.5112,
+      "step": 1563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000234,
+      "loss": 0.4109,
+      "step": 1564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023414999999999997,
+      "loss": 0.3154,
+      "step": 1565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023429999999999998,
+      "loss": 0.336,
+      "step": 1566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023444999999999998,
+      "loss": 0.3414,
+      "step": 1567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023459999999999998,
+      "loss": 0.3863,
+      "step": 1568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023474999999999996,
+      "loss": 0.5777,
+      "step": 1569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002349,
+      "loss": 0.5057,
+      "step": 1570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023504999999999997,
+      "loss": 0.2807,
+      "step": 1571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002352,
+      "loss": 0.2992,
+      "step": 1572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023534999999999997,
+      "loss": 0.3608,
+      "step": 1573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023549999999999998,
+      "loss": 0.4082,
+      "step": 1574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023564999999999998,
+      "loss": 0.5511,
+      "step": 1575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023579999999999999,
+      "loss": 0.4322,
+      "step": 1576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00023594999999999996,
+      "loss": 0.3247,
+      "step": 1577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002361,
+      "loss": 0.3612,
+      "step": 1578
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023624999999999997,
+      "loss": 0.3123,
+      "step": 1579
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002364,
+      "loss": 0.4227,
+      "step": 1580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023654999999999998,
+      "loss": 0.4734,
+      "step": 1581
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023669999999999998,
+      "loss": 0.483,
+      "step": 1582
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023684999999999998,
+      "loss": 0.3566,
+      "step": 1583
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000237,
+      "loss": 0.416,
+      "step": 1584
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023714999999999996,
+      "loss": 0.3711,
+      "step": 1585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002373,
+      "loss": 0.4601,
+      "step": 1586
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023744999999999997,
+      "loss": 0.4445,
+      "step": 1587
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002376,
+      "loss": 0.6517,
+      "step": 1588
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023774999999999998,
+      "loss": 0.4764,
+      "step": 1589
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023789999999999998,
+      "loss": 0.3553,
+      "step": 1590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023804999999999999,
+      "loss": 0.4063,
+      "step": 1591
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002382,
+      "loss": 0.4038,
+      "step": 1592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023834999999999997,
+      "loss": 0.4151,
+      "step": 1593
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002385,
+      "loss": 0.5916,
+      "step": 1594
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023864999999999997,
+      "loss": 0.3067,
+      "step": 1595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002388,
+      "loss": 0.5358,
+      "step": 1596
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023894999999999998,
+      "loss": 0.4249,
+      "step": 1597
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023909999999999998,
+      "loss": 0.3646,
+      "step": 1598
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023925,
+      "loss": 0.3942,
+      "step": 1599
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002394,
+      "loss": 0.5582,
+      "step": 1600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023954999999999997,
+      "loss": 0.3051,
+      "step": 1601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002397,
+      "loss": 0.4144,
+      "step": 1602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023984999999999998,
+      "loss": 0.3256,
+      "step": 1603
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.285,
+      "step": 1604
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024014999999999998,
+      "loss": 0.4479,
+      "step": 1605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024029999999999999,
+      "loss": 0.7009,
+      "step": 1606
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024044999999999996,
+      "loss": 0.7501,
+      "step": 1607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002406,
+      "loss": 0.46,
+      "step": 1608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024074999999999997,
+      "loss": 0.3018,
+      "step": 1609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002409,
+      "loss": 0.325,
+      "step": 1610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024104999999999998,
+      "loss": 0.3339,
+      "step": 1611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024119999999999998,
+      "loss": 0.401,
+      "step": 1612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024134999999999998,
+      "loss": 0.6745,
+      "step": 1613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002415,
+      "loss": 0.4913,
+      "step": 1614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024164999999999996,
+      "loss": 0.3301,
+      "step": 1615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002418,
+      "loss": 0.3095,
+      "step": 1616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024194999999999997,
+      "loss": 0.3842,
+      "step": 1617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002421,
+      "loss": 0.5149,
+      "step": 1618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024224999999999998,
+      "loss": 0.6425,
+      "step": 1619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024239999999999998,
+      "loss": 0.3871,
+      "step": 1620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024255,
+      "loss": 0.3703,
+      "step": 1621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002427,
+      "loss": 0.3391,
+      "step": 1622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024284999999999997,
+      "loss": 0.3327,
+      "step": 1623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000243,
+      "loss": 0.3773,
+      "step": 1624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024314999999999997,
+      "loss": 0.6078,
+      "step": 1625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002433,
+      "loss": 0.4332,
+      "step": 1626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024344999999999998,
+      "loss": 0.3157,
+      "step": 1627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024359999999999999,
+      "loss": 0.2755,
+      "step": 1628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024375,
+      "loss": 0.3527,
+      "step": 1629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00024389999999999997,
+      "loss": 0.4679,
+      "step": 1630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024404999999999997,
+      "loss": 0.3776,
+      "step": 1631
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024419999999999997,
+      "loss": 0.6829,
+      "step": 1632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024435,
+      "loss": 0.4308,
+      "step": 1633
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002445,
+      "loss": 0.3704,
+      "step": 1634
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024464999999999996,
+      "loss": 0.3575,
+      "step": 1635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002448,
+      "loss": 0.3722,
+      "step": 1636
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024494999999999996,
+      "loss": 0.4474,
+      "step": 1637
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024509999999999994,
+      "loss": 0.7114,
+      "step": 1638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024524999999999997,
+      "loss": 0.3221,
+      "step": 1639
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024539999999999995,
+      "loss": 0.3371,
+      "step": 1640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024555,
+      "loss": 0.3409,
+      "step": 1641
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024569999999999995,
+      "loss": 0.3339,
+      "step": 1642
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024585,
+      "loss": 0.4314,
+      "step": 1643
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 0.6105,
+      "step": 1644
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024615,
+      "loss": 0.3399,
+      "step": 1645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024629999999999997,
+      "loss": 0.4323,
+      "step": 1646
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024645,
+      "loss": 0.3212,
+      "step": 1647
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002466,
+      "loss": 0.5233,
+      "step": 1648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024675,
+      "loss": 0.4401,
+      "step": 1649
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002469,
+      "loss": 0.6842,
+      "step": 1650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024704999999999996,
+      "loss": 0.347,
+      "step": 1651
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002472,
+      "loss": 0.3729,
+      "step": 1652
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024734999999999997,
+      "loss": 0.3762,
+      "step": 1653
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024749999999999994,
+      "loss": 0.3736,
+      "step": 1654
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024765,
+      "loss": 0.4536,
+      "step": 1655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00024779999999999995,
+      "loss": 0.3947,
+      "step": 1656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024795,
+      "loss": 0.5152,
+      "step": 1657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024809999999999996,
+      "loss": 0.378,
+      "step": 1658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024825,
+      "loss": 0.4485,
+      "step": 1659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024839999999999997,
+      "loss": 0.4038,
+      "step": 1660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024855,
+      "loss": 0.2943,
+      "step": 1661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002487,
+      "loss": 0.4128,
+      "step": 1662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024885,
+      "loss": 0.6692,
+      "step": 1663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000249,
+      "loss": 0.399,
+      "step": 1664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024914999999999996,
+      "loss": 0.2748,
+      "step": 1665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002493,
+      "loss": 0.4149,
+      "step": 1666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024944999999999996,
+      "loss": 0.2814,
+      "step": 1667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024959999999999994,
+      "loss": 0.4528,
+      "step": 1668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024974999999999997,
+      "loss": 0.6662,
+      "step": 1669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00024989999999999995,
+      "loss": 0.2925,
+      "step": 1670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00025005,
+      "loss": 0.3395,
+      "step": 1671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00025019999999999996,
+      "loss": 0.3167,
+      "step": 1672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00025035,
+      "loss": 0.2626,
+      "step": 1673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00025049999999999996,
+      "loss": 0.4032,
+      "step": 1674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00025065,
+      "loss": 0.495,
+      "step": 1675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00025079999999999997,
+      "loss": 0.303,
+      "step": 1676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00025095,
+      "loss": 0.4268,
+      "step": 1677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002511,
+      "loss": 0.2754,
+      "step": 1678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00025125,
+      "loss": 0.4787,
+      "step": 1679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002514,
+      "loss": 0.3937,
+      "step": 1680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00025154999999999996,
+      "loss": 0.4137,
+      "step": 1681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002517,
+      "loss": 0.6336,
+      "step": 1682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025184999999999997,
+      "loss": 0.3977,
+      "step": 1683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 0.3135,
+      "step": 1684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025215,
+      "loss": 0.2999,
+      "step": 1685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025229999999999995,
+      "loss": 0.4743,
+      "step": 1686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025245,
+      "loss": 0.3828,
+      "step": 1687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025259999999999996,
+      "loss": 0.6419,
+      "step": 1688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025275,
+      "loss": 0.3069,
+      "step": 1689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025289999999999997,
+      "loss": 0.3387,
+      "step": 1690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025305,
+      "loss": 0.3232,
+      "step": 1691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002532,
+      "loss": 0.459,
+      "step": 1692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025335,
+      "loss": 0.3989,
+      "step": 1693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002535,
+      "loss": 0.7129,
+      "step": 1694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025365,
+      "loss": 0.3408,
+      "step": 1695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002538,
+      "loss": 0.2934,
+      "step": 1696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025394999999999997,
+      "loss": 0.3275,
+      "step": 1697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002541,
+      "loss": 0.3292,
+      "step": 1698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025425,
+      "loss": 0.6424,
+      "step": 1699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025439999999999995,
+      "loss": 0.5283,
+      "step": 1700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025455,
+      "loss": 0.3204,
+      "step": 1701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025469999999999996,
+      "loss": 0.4041,
+      "step": 1702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025485,
+      "loss": 0.3311,
+      "step": 1703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 0.3359,
+      "step": 1704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025515,
+      "loss": 0.3282,
+      "step": 1705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025529999999999997,
+      "loss": 0.4856,
+      "step": 1706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00025545,
+      "loss": 0.5226,
+      "step": 1707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002556,
+      "loss": 0.4158,
+      "step": 1708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025575,
+      "loss": 0.3266,
+      "step": 1709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002559,
+      "loss": 0.356,
+      "step": 1710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025604999999999996,
+      "loss": 0.399,
+      "step": 1711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002562,
+      "loss": 0.2848,
+      "step": 1712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025634999999999997,
+      "loss": 0.6188,
+      "step": 1713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025649999999999995,
+      "loss": 0.3489,
+      "step": 1714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025665,
+      "loss": 0.357,
+      "step": 1715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025679999999999995,
+      "loss": 0.275,
+      "step": 1716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025695,
+      "loss": 0.5307,
+      "step": 1717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025709999999999996,
+      "loss": 0.6004,
+      "step": 1718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025725,
+      "loss": 0.9478,
+      "step": 1719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025739999999999997,
+      "loss": 0.3939,
+      "step": 1720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025755,
+      "loss": 0.3036,
+      "step": 1721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002577,
+      "loss": 0.3687,
+      "step": 1722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025785,
+      "loss": 0.3219,
+      "step": 1723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000258,
+      "loss": 0.3616,
+      "step": 1724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025815,
+      "loss": 0.6523,
+      "step": 1725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002583,
+      "loss": 0.367,
+      "step": 1726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025844999999999997,
+      "loss": 0.5363,
+      "step": 1727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002586,
+      "loss": 0.3158,
+      "step": 1728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025875,
+      "loss": 0.3484,
+      "step": 1729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025889999999999995,
+      "loss": 0.4997,
+      "step": 1730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025905,
+      "loss": 0.4978,
+      "step": 1731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025919999999999996,
+      "loss": 0.667,
+      "step": 1732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025935,
+      "loss": 0.4227,
+      "step": 1733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00025949999999999997,
+      "loss": 0.2998,
+      "step": 1734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025965,
+      "loss": 0.4048,
+      "step": 1735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025979999999999997,
+      "loss": 0.486,
+      "step": 1736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025995,
+      "loss": 0.3948,
+      "step": 1737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002601,
+      "loss": 0.6103,
+      "step": 1738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026025,
+      "loss": 0.3665,
+      "step": 1739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002604,
+      "loss": 0.3455,
+      "step": 1740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026055,
+      "loss": 0.4215,
+      "step": 1741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002607,
+      "loss": 0.4263,
+      "step": 1742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026084999999999997,
+      "loss": 0.3698,
+      "step": 1743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000261,
+      "loss": 0.6104,
+      "step": 1744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026115,
+      "loss": 0.3683,
+      "step": 1745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026129999999999995,
+      "loss": 0.6167,
+      "step": 1746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026145,
+      "loss": 0.3349,
+      "step": 1747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026159999999999996,
+      "loss": 0.3633,
+      "step": 1748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026175,
+      "loss": 0.3913,
+      "step": 1749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026189999999999997,
+      "loss": 0.4326,
+      "step": 1750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026205,
+      "loss": 0.4648,
+      "step": 1751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002622,
+      "loss": 0.4772,
+      "step": 1752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026235,
+      "loss": 0.3741,
+      "step": 1753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002625,
+      "loss": 0.356,
+      "step": 1754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026264999999999996,
+      "loss": 0.3391,
+      "step": 1755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002628,
+      "loss": 0.5197,
+      "step": 1756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026294999999999997,
+      "loss": 0.6269,
+      "step": 1757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002631,
+      "loss": 0.4266,
+      "step": 1758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026325,
+      "loss": 0.4073,
+      "step": 1759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026339999999999995,
+      "loss": 0.3225,
+      "step": 1760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026355,
+      "loss": 0.3619,
+      "step": 1761
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026369999999999996,
+      "loss": 0.3425,
+      "step": 1762
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026384999999999994,
+      "loss": 0.663,
+      "step": 1763
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 0.4066,
+      "step": 1764
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026414999999999994,
+      "loss": 0.3361,
+      "step": 1765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002643,
+      "loss": 0.3924,
+      "step": 1766
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026444999999999995,
+      "loss": 0.3784,
+      "step": 1767
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002646,
+      "loss": 0.4516,
+      "step": 1768
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026474999999999996,
+      "loss": 0.5593,
+      "step": 1769
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002649,
+      "loss": 0.495,
+      "step": 1770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026504999999999996,
+      "loss": 0.3516,
+      "step": 1771
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002652,
+      "loss": 0.4727,
+      "step": 1772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026534999999999997,
+      "loss": 0.3253,
+      "step": 1773
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002655,
+      "loss": 0.4676,
+      "step": 1774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026565,
+      "loss": 0.7224,
+      "step": 1775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026579999999999996,
+      "loss": 0.3621,
+      "step": 1776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026595,
+      "loss": 0.3222,
+      "step": 1777
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026609999999999996,
+      "loss": 0.4181,
+      "step": 1778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026624999999999994,
+      "loss": 0.3882,
+      "step": 1779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026639999999999997,
+      "loss": 0.389,
+      "step": 1780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026654999999999995,
+      "loss": 0.3662,
+      "step": 1781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002667,
+      "loss": 0.7208,
+      "step": 1782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026684999999999995,
+      "loss": 0.3071,
+      "step": 1783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000267,
+      "loss": 0.4017,
+      "step": 1784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026714999999999996,
+      "loss": 0.2605,
+      "step": 1785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002673,
+      "loss": 0.2418,
+      "step": 1786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00026744999999999997,
+      "loss": 0.3153,
+      "step": 1787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002676,
+      "loss": 0.6237,
+      "step": 1788
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00026775,
+      "loss": 1.0159,
+      "step": 1789
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002679,
+      "loss": 0.2922,
+      "step": 1790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00026805,
+      "loss": 0.4335,
+      "step": 1791
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00026819999999999996,
+      "loss": 0.3023,
+      "step": 1792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00026835,
+      "loss": 0.3996,
+      "step": 1793
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00026849999999999997,
+      "loss": 0.8867,
+      "step": 1794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00026864999999999994,
+      "loss": 0.448,
+      "step": 1795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002688,
+      "loss": 0.3295,
+      "step": 1796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00026894999999999995,
+      "loss": 0.2998,
+      "step": 1797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002691,
+      "loss": 0.5078,
+      "step": 1798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00026924999999999996,
+      "loss": 0.279,
+      "step": 1799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002694,
+      "loss": 0.6901,
+      "step": 1800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00026954999999999997,
+      "loss": 0.4058,
+      "step": 1801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002697,
+      "loss": 0.3048,
+      "step": 1802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00026984999999999997,
+      "loss": 0.2699,
+      "step": 1803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00027,
+      "loss": 0.5064,
+      "step": 1804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00027015,
+      "loss": 0.6655,
+      "step": 1805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00027029999999999996,
+      "loss": 0.4683,
+      "step": 1806
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00027045,
+      "loss": 0.4415,
+      "step": 1807
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00027059999999999996,
+      "loss": 0.2806,
+      "step": 1808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00027074999999999994,
+      "loss": 0.2925,
+      "step": 1809
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00027089999999999997,
+      "loss": 0.2438,
+      "step": 1810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00027104999999999995,
+      "loss": 0.329,
+      "step": 1811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002712,
+      "loss": 0.3345,
+      "step": 1812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00027134999999999995,
+      "loss": 0.6175,
+      "step": 1813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002715,
+      "loss": 0.3287,
+      "step": 1814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027164999999999996,
+      "loss": 0.3496,
+      "step": 1815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002718,
+      "loss": 0.3495,
+      "step": 1816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027194999999999997,
+      "loss": 0.3575,
+      "step": 1817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002721,
+      "loss": 0.5037,
+      "step": 1818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027225,
+      "loss": 0.4339,
+      "step": 1819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002724,
+      "loss": 0.3281,
+      "step": 1820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027255,
+      "loss": 0.3667,
+      "step": 1821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027269999999999996,
+      "loss": 0.3467,
+      "step": 1822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027285,
+      "loss": 0.3212,
+      "step": 1823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 0.3985,
+      "step": 1824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027314999999999994,
+      "loss": 0.4596,
+      "step": 1825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002733,
+      "loss": 0.3693,
+      "step": 1826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027344999999999995,
+      "loss": 0.287,
+      "step": 1827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002736,
+      "loss": 0.2869,
+      "step": 1828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027374999999999996,
+      "loss": 0.3756,
+      "step": 1829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002739,
+      "loss": 0.4039,
+      "step": 1830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027404999999999997,
+      "loss": 0.3759,
+      "step": 1831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002742,
+      "loss": 0.5336,
+      "step": 1832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027435,
+      "loss": 0.3946,
+      "step": 1833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002745,
+      "loss": 0.3581,
+      "step": 1834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027465,
+      "loss": 0.3379,
+      "step": 1835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002748,
+      "loss": 0.3663,
+      "step": 1836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027495,
+      "loss": 0.3852,
+      "step": 1837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027509999999999996,
+      "loss": 0.4843,
+      "step": 1838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00027525,
+      "loss": 0.2874,
+      "step": 1839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 0.3893,
+      "step": 1840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027554999999999995,
+      "loss": 0.2802,
+      "step": 1841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002757,
+      "loss": 0.3652,
+      "step": 1842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027584999999999996,
+      "loss": 0.4906,
+      "step": 1843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000276,
+      "loss": 0.5078,
+      "step": 1844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027614999999999996,
+      "loss": 0.3416,
+      "step": 1845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002763,
+      "loss": 0.3793,
+      "step": 1846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027644999999999997,
+      "loss": 0.2932,
+      "step": 1847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002766,
+      "loss": 0.2694,
+      "step": 1848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027675,
+      "loss": 0.4693,
+      "step": 1849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002769,
+      "loss": 0.5572,
+      "step": 1850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027705,
+      "loss": 0.282,
+      "step": 1851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002772,
+      "loss": 0.3416,
+      "step": 1852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027735,
+      "loss": 0.4358,
+      "step": 1853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027749999999999997,
+      "loss": 0.314,
+      "step": 1854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027764999999999995,
+      "loss": 0.2929,
+      "step": 1855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002778,
+      "loss": 0.4677,
+      "step": 1856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027794999999999995,
+      "loss": 0.4993,
+      "step": 1857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002781,
+      "loss": 0.397,
+      "step": 1858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027824999999999996,
+      "loss": 0.3003,
+      "step": 1859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002784,
+      "loss": 0.2987,
+      "step": 1860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027854999999999997,
+      "loss": 0.2838,
+      "step": 1861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002787,
+      "loss": 0.312,
+      "step": 1862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027885,
+      "loss": 0.4438,
+      "step": 1863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000279,
+      "loss": 0.2737,
+      "step": 1864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00027915,
+      "loss": 0.5436,
+      "step": 1865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002793,
+      "loss": 0.2572,
+      "step": 1866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00027945,
+      "loss": 0.2483,
+      "step": 1867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00027959999999999997,
+      "loss": 0.4455,
+      "step": 1868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00027975,
+      "loss": 0.7475,
+      "step": 1869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002799,
+      "loss": 0.3748,
+      "step": 1870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028004999999999995,
+      "loss": 0.5246,
+      "step": 1871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002802,
+      "loss": 0.3339,
+      "step": 1872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028034999999999996,
+      "loss": 0.3398,
+      "step": 1873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002805,
+      "loss": 0.4235,
+      "step": 1874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028064999999999996,
+      "loss": 0.629,
+      "step": 1875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002808,
+      "loss": 0.3983,
+      "step": 1876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028094999999999997,
+      "loss": 0.3743,
+      "step": 1877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002811,
+      "loss": 0.3094,
+      "step": 1878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028125,
+      "loss": 0.3457,
+      "step": 1879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028139999999999996,
+      "loss": 0.3468,
+      "step": 1880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028155,
+      "loss": 0.4321,
+      "step": 1881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028169999999999996,
+      "loss": 0.667,
+      "step": 1882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028185,
+      "loss": 0.3535,
+      "step": 1883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 0.368,
+      "step": 1884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028215,
+      "loss": 0.2782,
+      "step": 1885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002823,
+      "loss": 0.3509,
+      "step": 1886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028244999999999995,
+      "loss": 0.3982,
+      "step": 1887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002826,
+      "loss": 0.627,
+      "step": 1888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028274999999999996,
+      "loss": 0.2877,
+      "step": 1889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028289999999999994,
+      "loss": 0.2544,
+      "step": 1890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028304999999999997,
+      "loss": 0.2848,
+      "step": 1891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028319999999999994,
+      "loss": 0.3243,
+      "step": 1892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028335,
+      "loss": 0.4344,
+      "step": 1893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028349999999999995,
+      "loss": 0.6745,
+      "step": 1894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028365,
+      "loss": 0.3554,
+      "step": 1895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028379999999999996,
+      "loss": 0.2719,
+      "step": 1896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028395,
+      "loss": 0.4391,
+      "step": 1897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028409999999999997,
+      "loss": 0.354,
+      "step": 1898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028425,
+      "loss": 0.3295,
+      "step": 1899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002844,
+      "loss": 0.6761,
+      "step": 1900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028455,
+      "loss": 0.5548,
+      "step": 1901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002847,
+      "loss": 0.3149,
+      "step": 1902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028484999999999996,
+      "loss": 0.2963,
+      "step": 1903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000285,
+      "loss": 0.3734,
+      "step": 1904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028514999999999997,
+      "loss": 0.4355,
+      "step": 1905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028529999999999994,
+      "loss": 0.4699,
+      "step": 1906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028544999999999997,
+      "loss": 0.5423,
+      "step": 1907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028559999999999995,
+      "loss": 0.3908,
+      "step": 1908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028575,
+      "loss": 0.2719,
+      "step": 1909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028589999999999996,
+      "loss": 0.2907,
+      "step": 1910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028605,
+      "loss": 0.3003,
+      "step": 1911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028619999999999996,
+      "loss": 0.3214,
+      "step": 1912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028635,
+      "loss": 0.5298,
+      "step": 1913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028649999999999997,
+      "loss": 0.3679,
+      "step": 1914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028665,
+      "loss": 0.4859,
+      "step": 1915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002868,
+      "loss": 0.2717,
+      "step": 1916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028694999999999995,
+      "loss": 0.4025,
+      "step": 1917
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002871,
+      "loss": 0.3051,
+      "step": 1918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028724999999999996,
+      "loss": 0.6741,
+      "step": 1919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028739999999999994,
+      "loss": 0.3428,
+      "step": 1920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028754999999999997,
+      "loss": 0.3598,
+      "step": 1921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028769999999999995,
+      "loss": 0.4819,
+      "step": 1922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028785,
+      "loss": 0.3501,
+      "step": 1923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 0.3576,
+      "step": 1924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028815,
+      "loss": 0.5136,
+      "step": 1925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028829999999999996,
+      "loss": 0.3086,
+      "step": 1926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028845,
+      "loss": 0.471,
+      "step": 1927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028859999999999997,
+      "loss": 0.2994,
+      "step": 1928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028875,
+      "loss": 0.2768,
+      "step": 1929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002889,
+      "loss": 0.4909,
+      "step": 1930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028905,
+      "loss": 0.4676,
+      "step": 1931
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002892,
+      "loss": 0.5044,
+      "step": 1932
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028934999999999996,
+      "loss": 0.3498,
+      "step": 1933
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002895,
+      "loss": 0.3382,
+      "step": 1934
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028964999999999997,
+      "loss": 0.4345,
+      "step": 1935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028979999999999994,
+      "loss": 0.2647,
+      "step": 1936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028995,
+      "loss": 0.4107,
+      "step": 1937
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029009999999999995,
+      "loss": 0.5284,
+      "step": 1938
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029025,
+      "loss": 0.2658,
+      "step": 1939
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029039999999999996,
+      "loss": 0.3359,
+      "step": 1940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029055,
+      "loss": 0.3547,
+      "step": 1941
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029069999999999996,
+      "loss": 0.3532,
+      "step": 1942
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029085,
+      "loss": 0.3523,
+      "step": 1943
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 0.4923,
+      "step": 1944
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029115,
+      "loss": 0.3318,
+      "step": 1945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002913,
+      "loss": 0.3174,
+      "step": 1946
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029145,
+      "loss": 0.3042,
+      "step": 1947
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002916,
+      "loss": 0.4115,
+      "step": 1948
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029174999999999996,
+      "loss": 0.4387,
+      "step": 1949
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002919,
+      "loss": 0.7413,
+      "step": 1950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029204999999999997,
+      "loss": 0.335,
+      "step": 1951
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029219999999999995,
+      "loss": 0.3781,
+      "step": 1952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029235,
+      "loss": 0.371,
+      "step": 1953
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029249999999999995,
+      "loss": 0.3176,
+      "step": 1954
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029265,
+      "loss": 0.5342,
+      "step": 1955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029279999999999996,
+      "loss": 0.4736,
+      "step": 1956
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029295,
+      "loss": 0.5081,
+      "step": 1957
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029309999999999997,
+      "loss": 0.3376,
+      "step": 1958
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029325,
+      "loss": 0.2966,
+      "step": 1959
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002934,
+      "loss": 0.2458,
+      "step": 1960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029355,
+      "loss": 0.3434,
+      "step": 1961
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002937,
+      "loss": 0.3517,
+      "step": 1962
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029384999999999996,
+      "loss": 0.7423,
+      "step": 1963
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000294,
+      "loss": 0.3797,
+      "step": 1964
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029414999999999997,
+      "loss": 0.2695,
+      "step": 1965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029429999999999994,
+      "loss": 0.3296,
+      "step": 1966
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029445,
+      "loss": 0.3245,
+      "step": 1967
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 0.4714,
+      "step": 1968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00029475,
+      "loss": 0.675,
+      "step": 1969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029489999999999996,
+      "loss": 0.4012,
+      "step": 1970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029505,
+      "loss": 0.4902,
+      "step": 1971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 0.2903,
+      "step": 1972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029535,
+      "loss": 0.2978,
+      "step": 1973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029549999999999997,
+      "loss": 0.4064,
+      "step": 1974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029565,
+      "loss": 0.8423,
+      "step": 1975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002958,
+      "loss": 0.3312,
+      "step": 1976
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029595,
+      "loss": 0.3453,
+      "step": 1977
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002961,
+      "loss": 0.2267,
+      "step": 1978
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029624999999999996,
+      "loss": 0.3051,
+      "step": 1979
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002964,
+      "loss": 0.3204,
+      "step": 1980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029654999999999997,
+      "loss": 0.3989,
+      "step": 1981
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029669999999999995,
+      "loss": 0.5797,
+      "step": 1982
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029685,
+      "loss": 0.3329,
+      "step": 1983
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 0.3404,
+      "step": 1984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029715,
+      "loss": 0.3242,
+      "step": 1985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029729999999999996,
+      "loss": 0.4122,
+      "step": 1986
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029745,
+      "loss": 0.3126,
+      "step": 1987
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 0.4817,
+      "step": 1988
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029775,
+      "loss": 0.2762,
+      "step": 1989
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002979,
+      "loss": 0.3891,
+      "step": 1990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029805,
+      "loss": 0.2793,
+      "step": 1991
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002982,
+      "loss": 0.3647,
+      "step": 1992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029835,
+      "loss": 0.3532,
+      "step": 1993
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002985,
+      "loss": 0.538,
+      "step": 1994
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029864999999999997,
+      "loss": 0.3403,
+      "step": 1995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002988,
+      "loss": 0.368,
+      "step": 1996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029895,
+      "loss": 0.456,
+      "step": 1997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029909999999999995,
+      "loss": 0.3681,
+      "step": 1998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029925,
+      "loss": 0.3906,
+      "step": 1999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029939999999999996,
+      "loss": 0.5641,
+      "step": 2000
+    },
+    {
+      "epoch": 0.77,
+      "eval_cer": 0.1283649062567064,
+      "eval_loss": 0.44599324464797974,
+      "eval_runtime": 46.817,
+      "eval_samples_per_second": 37.444,
+      "eval_steps_per_second": 4.699,
+      "eval_wer": 0.2540163823143413,
+      "step": 2000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029955,
+      "loss": 0.3193,
+      "step": 2001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029969999999999997,
+      "loss": 0.2712,
+      "step": 2002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029985,
+      "loss": 0.2605,
+      "step": 2003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0003,
+      "loss": 0.2373,
+      "step": 2004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002999875466998754,
+      "loss": 0.3615,
+      "step": 2005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002999750933997509,
+      "loss": 0.3647,
+      "step": 2006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002999626400996264,
+      "loss": 0.4082,
+      "step": 2007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029995018679950186,
+      "loss": 0.3533,
+      "step": 2008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002999377334993773,
+      "loss": 0.3825,
+      "step": 2009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002999252801992528,
+      "loss": 0.2615,
+      "step": 2010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002999128268991282,
+      "loss": 0.3789,
+      "step": 2011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002999003735990037,
+      "loss": 0.3202,
+      "step": 2012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002998879202988792,
+      "loss": 0.7153,
+      "step": 2013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002998754669987547,
+      "loss": 0.3097,
+      "step": 2014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002998630136986301,
+      "loss": 0.4163,
+      "step": 2015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002998505603985056,
+      "loss": 0.314,
+      "step": 2016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000299838107098381,
+      "loss": 0.2576,
+      "step": 2017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002998256537982565,
+      "loss": 0.494,
+      "step": 2018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000299813200498132,
+      "loss": 0.4785,
+      "step": 2019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029980074719800743,
+      "loss": 0.3612,
+      "step": 2020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002997882938978829,
+      "loss": 0.3943,
+      "step": 2021
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029977584059775835,
+      "loss": 0.3607,
+      "step": 2022
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029976338729763383,
+      "loss": 0.2729,
+      "step": 2023
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002997509339975093,
+      "loss": 0.4042,
+      "step": 2024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002997384806973848,
+      "loss": 0.51,
+      "step": 2025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029972602739726024,
+      "loss": 0.3098,
+      "step": 2026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002997135740971357,
+      "loss": 0.3043,
+      "step": 2027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029970112079701116,
+      "loss": 0.2955,
+      "step": 2028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029968866749688664,
+      "loss": 0.2742,
+      "step": 2029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029967621419676213,
+      "loss": 0.3005,
+      "step": 2030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002996637608966376,
+      "loss": 0.3693,
+      "step": 2031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029965130759651305,
+      "loss": 0.7927,
+      "step": 2032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029963885429638853,
+      "loss": 0.3069,
+      "step": 2033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029962640099626397,
+      "loss": 0.3552,
+      "step": 2034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029961394769613945,
+      "loss": 0.3127,
+      "step": 2035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029960149439601494,
+      "loss": 0.301,
+      "step": 2036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029958904109589037,
+      "loss": 0.5081,
+      "step": 2037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029957658779576586,
+      "loss": 0.5826,
+      "step": 2038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002995641344956413,
+      "loss": 0.4267,
+      "step": 2039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002995516811955168,
+      "loss": 0.4755,
+      "step": 2040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029953922789539226,
+      "loss": 0.2839,
+      "step": 2041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029952677459526775,
+      "loss": 0.3348,
+      "step": 2042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002995143212951432,
+      "loss": 0.273,
+      "step": 2043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029950186799501867,
+      "loss": 0.5651,
+      "step": 2044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002994894146948941,
+      "loss": 0.3016,
+      "step": 2045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002994769613947696,
+      "loss": 0.3531,
+      "step": 2046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029946450809464507,
+      "loss": 0.3328,
+      "step": 2047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00029945205479452056,
+      "loss": 0.29,
+      "step": 2048
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000299439601494396,
+      "loss": 0.3967,
+      "step": 2049
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002994271481942715,
+      "loss": 0.4981,
+      "step": 2050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002994146948941469,
+      "loss": 0.3592,
+      "step": 2051
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002994022415940224,
+      "loss": 0.3873,
+      "step": 2052
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002993897882938979,
+      "loss": 0.3991,
+      "step": 2053
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002993773349937733,
+      "loss": 0.2775,
+      "step": 2054
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002993648816936488,
+      "loss": 0.3847,
+      "step": 2055
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029935242839352423,
+      "loss": 0.5031,
+      "step": 2056
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002993399750933997,
+      "loss": 0.4656,
+      "step": 2057
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002993275217932752,
+      "loss": 0.3242,
+      "step": 2058
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002993150684931507,
+      "loss": 0.4211,
+      "step": 2059
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002993026151930261,
+      "loss": 0.3812,
+      "step": 2060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002992901618929016,
+      "loss": 0.2921,
+      "step": 2061
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029927770859277704,
+      "loss": 0.5213,
+      "step": 2062
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002992652552926525,
+      "loss": 0.5365,
+      "step": 2063
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000299252801992528,
+      "loss": 0.3748,
+      "step": 2064
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002992403486924035,
+      "loss": 0.3187,
+      "step": 2065
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029922789539227893,
+      "loss": 0.5432,
+      "step": 2066
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029921544209215436,
+      "loss": 0.3527,
+      "step": 2067
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029920298879202985,
+      "loss": 0.4966,
+      "step": 2068
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029919053549190533,
+      "loss": 0.5426,
+      "step": 2069
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002991780821917808,
+      "loss": 0.2777,
+      "step": 2070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029916562889165625,
+      "loss": 0.302,
+      "step": 2071
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029915317559153174,
+      "loss": 0.3697,
+      "step": 2072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029914072229140717,
+      "loss": 0.3005,
+      "step": 2073
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029912826899128266,
+      "loss": 0.4073,
+      "step": 2074
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029911581569115814,
+      "loss": 0.8599,
+      "step": 2075
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029910336239103363,
+      "loss": 0.5463,
+      "step": 2076
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029909090909090906,
+      "loss": 0.3145,
+      "step": 2077
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029907845579078455,
+      "loss": 0.338,
+      "step": 2078
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029906600249066,
+      "loss": 0.3332,
+      "step": 2079
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029905354919053547,
+      "loss": 0.2777,
+      "step": 2080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029904109589041095,
+      "loss": 0.5309,
+      "step": 2081
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029902864259028644,
+      "loss": 0.5565,
+      "step": 2082
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029901618929016187,
+      "loss": 0.2682,
+      "step": 2083
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002990037359900373,
+      "loss": 0.2583,
+      "step": 2084
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002989912826899128,
+      "loss": 0.6114,
+      "step": 2085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002989788293897883,
+      "loss": 0.4544,
+      "step": 2086
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029896637608966376,
+      "loss": 0.4325,
+      "step": 2087
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002989539227895392,
+      "loss": 0.601,
+      "step": 2088
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002989414694894147,
+      "loss": 0.43,
+      "step": 2089
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002989290161892901,
+      "loss": 0.2754,
+      "step": 2090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002989165628891656,
+      "loss": 0.3011,
+      "step": 2091
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002989041095890411,
+      "loss": 0.4705,
+      "step": 2092
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029889165628891657,
+      "loss": 0.4048,
+      "step": 2093
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000298879202988792,
+      "loss": 0.5185,
+      "step": 2094
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002988667496886675,
+      "loss": 0.516,
+      "step": 2095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002988542963885429,
+      "loss": 0.3392,
+      "step": 2096
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002988418430884184,
+      "loss": 0.3248,
+      "step": 2097
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002988293897882939,
+      "loss": 0.324,
+      "step": 2098
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002988169364881693,
+      "loss": 0.3184,
+      "step": 2099
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002988044831880448,
+      "loss": 0.5746,
+      "step": 2100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029879202988792024,
+      "loss": 0.5482,
+      "step": 2101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029877957658779573,
+      "loss": 0.3178,
+      "step": 2102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002987671232876712,
+      "loss": 0.3341,
+      "step": 2103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002987546699875467,
+      "loss": 0.447,
+      "step": 2104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029874221668742214,
+      "loss": 0.4106,
+      "step": 2105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002987297633872976,
+      "loss": 0.5751,
+      "step": 2106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029871731008717305,
+      "loss": 0.6183,
+      "step": 2107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029870485678704854,
+      "loss": 0.3539,
+      "step": 2108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000298692403486924,
+      "loss": 0.4863,
+      "step": 2109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002986799501867995,
+      "loss": 0.2826,
+      "step": 2110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029866749688667495,
+      "loss": 0.3059,
+      "step": 2111
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029865504358655043,
+      "loss": 0.4184,
+      "step": 2112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029864259028642586,
+      "loss": 0.4832,
+      "step": 2113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029863013698630135,
+      "loss": 0.3106,
+      "step": 2114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029861768368617684,
+      "loss": 0.3343,
+      "step": 2115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029860523038605227,
+      "loss": 0.2665,
+      "step": 2116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029859277708592775,
+      "loss": 0.3139,
+      "step": 2117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002985803237858032,
+      "loss": 0.3571,
+      "step": 2118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029856787048567867,
+      "loss": 0.5099,
+      "step": 2119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029855541718555416,
+      "loss": 0.4976,
+      "step": 2120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029854296388542965,
+      "loss": 0.382,
+      "step": 2121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002985305105853051,
+      "loss": 0.3341,
+      "step": 2122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029851805728518056,
+      "loss": 0.326,
+      "step": 2123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000298505603985056,
+      "loss": 0.3243,
+      "step": 2124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002984931506849315,
+      "loss": 0.4937,
+      "step": 2125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029848069738480697,
+      "loss": 0.3155,
+      "step": 2126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029846824408468245,
+      "loss": 0.3575,
+      "step": 2127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002984557907845579,
+      "loss": 0.3211,
+      "step": 2128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029844333748443337,
+      "loss": 0.5386,
+      "step": 2129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002984308841843088,
+      "loss": 0.2793,
+      "step": 2130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002984184308841843,
+      "loss": 0.403,
+      "step": 2131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002984059775840598,
+      "loss": 0.7352,
+      "step": 2132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002983935242839352,
+      "loss": 0.38,
+      "step": 2133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002983810709838107,
+      "loss": 0.3899,
+      "step": 2134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029836861768368613,
+      "loss": 0.3087,
+      "step": 2135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002983561643835616,
+      "loss": 0.2631,
+      "step": 2136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002983437110834371,
+      "loss": 0.492,
+      "step": 2137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002983312577833126,
+      "loss": 0.4174,
+      "step": 2138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000298318804483188,
+      "loss": 0.3541,
+      "step": 2139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002983063511830635,
+      "loss": 0.2503,
+      "step": 2140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029829389788293894,
+      "loss": 0.3447,
+      "step": 2141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002982814445828144,
+      "loss": 0.2808,
+      "step": 2142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002982689912826899,
+      "loss": 0.4202,
+      "step": 2143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002982565379825654,
+      "loss": 0.5355,
+      "step": 2144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029824408468244083,
+      "loss": 0.4301,
+      "step": 2145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029823163138231626,
+      "loss": 0.3099,
+      "step": 2146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029821917808219175,
+      "loss": 0.3675,
+      "step": 2147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029820672478206723,
+      "loss": 0.2655,
+      "step": 2148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002981942714819427,
+      "loss": 0.4656,
+      "step": 2149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029818181818181815,
+      "loss": 0.5101,
+      "step": 2150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029816936488169364,
+      "loss": 0.3398,
+      "step": 2151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00029815691158156907,
+      "loss": 0.2669,
+      "step": 2152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029814445828144456,
+      "loss": 0.241,
+      "step": 2153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029813200498132004,
+      "loss": 0.286,
+      "step": 2154
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029811955168119553,
+      "loss": 0.3917,
+      "step": 2155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029810709838107096,
+      "loss": 0.3947,
+      "step": 2156
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029809464508094645,
+      "loss": 0.483,
+      "step": 2157
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002980821917808219,
+      "loss": 0.3074,
+      "step": 2158
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029806973848069736,
+      "loss": 0.4348,
+      "step": 2159
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029805728518057285,
+      "loss": 0.3652,
+      "step": 2160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002980448318804483,
+      "loss": 0.2356,
+      "step": 2161
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029803237858032377,
+      "loss": 0.2798,
+      "step": 2162
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002980199252801992,
+      "loss": 0.6888,
+      "step": 2163
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002980074719800747,
+      "loss": 0.3042,
+      "step": 2164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002979950186799502,
+      "loss": 0.2678,
+      "step": 2165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002979825653798256,
+      "loss": 0.2448,
+      "step": 2166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002979701120797011,
+      "loss": 0.3657,
+      "step": 2167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002979576587795766,
+      "loss": 0.3435,
+      "step": 2168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000297945205479452,
+      "loss": 0.6969,
+      "step": 2169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002979327521793275,
+      "loss": 0.2668,
+      "step": 2170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000297920298879203,
+      "loss": 0.3937,
+      "step": 2171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002979078455790784,
+      "loss": 0.2786,
+      "step": 2172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002978953922789539,
+      "loss": 0.3075,
+      "step": 2173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002978829389788294,
+      "loss": 0.3602,
+      "step": 2174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002978704856787048,
+      "loss": 0.8382,
+      "step": 2175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002978580323785803,
+      "loss": 0.3885,
+      "step": 2176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029784557907845574,
+      "loss": 0.2798,
+      "step": 2177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002978331257783312,
+      "loss": 0.3202,
+      "step": 2178
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002978206724782067,
+      "loss": 0.3627,
+      "step": 2179
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029780821917808214,
+      "loss": 0.4097,
+      "step": 2180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029779576587795763,
+      "loss": 0.4041,
+      "step": 2181
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002977833125778331,
+      "loss": 0.4263,
+      "step": 2182
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029777085927770855,
+      "loss": 0.4016,
+      "step": 2183
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029775840597758403,
+      "loss": 0.2748,
+      "step": 2184
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002977459526774595,
+      "loss": 0.2903,
+      "step": 2185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029773349937733495,
+      "loss": 0.3558,
+      "step": 2186
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029772104607721044,
+      "loss": 0.3131,
+      "step": 2187
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029770859277708587,
+      "loss": 0.6664,
+      "step": 2188
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029769613947696136,
+      "loss": 0.3479,
+      "step": 2189
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029768368617683684,
+      "loss": 0.26,
+      "step": 2190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029767123287671233,
+      "loss": 0.257,
+      "step": 2191
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029765877957658776,
+      "loss": 0.305,
+      "step": 2192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029764632627646325,
+      "loss": 0.2484,
+      "step": 2193
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002976338729763387,
+      "loss": 0.5059,
+      "step": 2194
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029762141967621417,
+      "loss": 1.2781,
+      "step": 2195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029760896637608965,
+      "loss": 0.3522,
+      "step": 2196
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002975965130759651,
+      "loss": 0.3089,
+      "step": 2197
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029758405977584057,
+      "loss": 0.4459,
+      "step": 2198
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000297571606475716,
+      "loss": 0.342,
+      "step": 2199
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002975591531755915,
+      "loss": 0.6177,
+      "step": 2200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000297546699875467,
+      "loss": 0.3439,
+      "step": 2201
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029753424657534246,
+      "loss": 0.2752,
+      "step": 2202
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002975217932752179,
+      "loss": 0.311,
+      "step": 2203
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002975093399750934,
+      "loss": 0.3689,
+      "step": 2204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002974968866749688,
+      "loss": 0.3377,
+      "step": 2205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002974844333748443,
+      "loss": 0.3155,
+      "step": 2206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002974719800747198,
+      "loss": 0.5546,
+      "step": 2207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029745952677459527,
+      "loss": 0.2786,
+      "step": 2208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002974470734744707,
+      "loss": 0.2804,
+      "step": 2209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029743462017434613,
+      "loss": 0.3783,
+      "step": 2210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002974221668742216,
+      "loss": 0.3642,
+      "step": 2211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002974097135740971,
+      "loss": 0.3127,
+      "step": 2212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002973972602739726,
+      "loss": 0.6511,
+      "step": 2213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000297384806973848,
+      "loss": 0.3001,
+      "step": 2214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002973723536737235,
+      "loss": 0.2604,
+      "step": 2215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029735990037359894,
+      "loss": 0.2511,
+      "step": 2216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029734744707347443,
+      "loss": 0.3651,
+      "step": 2217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002973349937733499,
+      "loss": 0.4288,
+      "step": 2218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002973225404732254,
+      "loss": 0.6166,
+      "step": 2219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029731008717310083,
+      "loss": 0.3148,
+      "step": 2220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002972976338729763,
+      "loss": 0.3496,
+      "step": 2221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029728518057285175,
+      "loss": 0.4589,
+      "step": 2222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029727272727272724,
+      "loss": 0.2815,
+      "step": 2223
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002972602739726027,
+      "loss": 0.2864,
+      "step": 2224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029724782067247816,
+      "loss": 0.4581,
+      "step": 2225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029723536737235364,
+      "loss": 0.3808,
+      "step": 2226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002972229140722291,
+      "loss": 0.7624,
+      "step": 2227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029721046077210456,
+      "loss": 0.2902,
+      "step": 2228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029719800747198005,
+      "loss": 0.2842,
+      "step": 2229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00029718555417185553,
+      "loss": 0.4757,
+      "step": 2230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029717310087173097,
+      "loss": 0.4046,
+      "step": 2231
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029716064757160645,
+      "loss": 0.4975,
+      "step": 2232
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002971481942714819,
+      "loss": 0.2656,
+      "step": 2233
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029713574097135737,
+      "loss": 0.6618,
+      "step": 2234
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029712328767123286,
+      "loss": 0.438,
+      "step": 2235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029711083437110834,
+      "loss": 0.3127,
+      "step": 2236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002970983810709838,
+      "loss": 0.2788,
+      "step": 2237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029708592777085926,
+      "loss": 0.527,
+      "step": 2238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002970734744707347,
+      "loss": 0.3812,
+      "step": 2239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002970610211706102,
+      "loss": 0.289,
+      "step": 2240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029704856787048567,
+      "loss": 0.2841,
+      "step": 2241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002970361145703611,
+      "loss": 0.3032,
+      "step": 2242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002970236612702366,
+      "loss": 0.3416,
+      "step": 2243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000297011207970112,
+      "loss": 0.6695,
+      "step": 2244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002969987546699875,
+      "loss": 0.2593,
+      "step": 2245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000296986301369863,
+      "loss": 0.2822,
+      "step": 2246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002969738480697385,
+      "loss": 0.2723,
+      "step": 2247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002969613947696139,
+      "loss": 0.2515,
+      "step": 2248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002969489414694894,
+      "loss": 0.2814,
+      "step": 2249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002969364881693648,
+      "loss": 0.4264,
+      "step": 2250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002969240348692403,
+      "loss": 0.4618,
+      "step": 2251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002969115815691158,
+      "loss": 0.3068,
+      "step": 2252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002968991282689913,
+      "loss": 0.4321,
+      "step": 2253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002968866749688667,
+      "loss": 0.2859,
+      "step": 2254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002968742216687422,
+      "loss": 0.3766,
+      "step": 2255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00029686176836861764,
+      "loss": 0.5108,
+      "step": 2256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002968493150684931,
+      "loss": 0.4047,
+      "step": 2257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002968368617683686,
+      "loss": 0.3443,
+      "step": 2258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029682440846824404,
+      "loss": 0.2666,
+      "step": 2259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002968119551681195,
+      "loss": 0.2827,
+      "step": 2260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029679950186799496,
+      "loss": 0.2694,
+      "step": 2261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029678704856787045,
+      "loss": 0.3935,
+      "step": 2262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029677459526774593,
+      "loss": 0.4644,
+      "step": 2263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002967621419676214,
+      "loss": 0.3754,
+      "step": 2264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029674968866749685,
+      "loss": 0.4417,
+      "step": 2265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029673723536737234,
+      "loss": 0.264,
+      "step": 2266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029672478206724777,
+      "loss": 0.4654,
+      "step": 2267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029671232876712325,
+      "loss": 0.3338,
+      "step": 2268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029669987546699874,
+      "loss": 0.5084,
+      "step": 2269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002966874221668742,
+      "loss": 0.3587,
+      "step": 2270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029667496886674966,
+      "loss": 0.2856,
+      "step": 2271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002966625155666251,
+      "loss": 0.32,
+      "step": 2272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002966500622665006,
+      "loss": 0.4161,
+      "step": 2273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029663760896637606,
+      "loss": 0.3451,
+      "step": 2274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029662515566625155,
+      "loss": 0.3911,
+      "step": 2275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000296612702366127,
+      "loss": 0.3382,
+      "step": 2276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029660024906600247,
+      "loss": 0.3617,
+      "step": 2277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002965877957658779,
+      "loss": 0.2655,
+      "step": 2278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002965753424657534,
+      "loss": 0.4064,
+      "step": 2279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029656288916562887,
+      "loss": 0.4847,
+      "step": 2280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029655043586550436,
+      "loss": 0.4491,
+      "step": 2281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002965379825653798,
+      "loss": 0.4549,
+      "step": 2282
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002965255292652553,
+      "loss": 0.2835,
+      "step": 2283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002965130759651307,
+      "loss": 0.2934,
+      "step": 2284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002965006226650062,
+      "loss": 0.3706,
+      "step": 2285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002964881693648817,
+      "loss": 0.4266,
+      "step": 2286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002964757160647571,
+      "loss": 0.2917,
+      "step": 2287
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002964632627646326,
+      "loss": 0.5799,
+      "step": 2288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029645080946450803,
+      "loss": 0.2808,
+      "step": 2289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002964383561643835,
+      "loss": 0.3272,
+      "step": 2290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000296425902864259,
+      "loss": 0.374,
+      "step": 2291
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002964134495641345,
+      "loss": 0.338,
+      "step": 2292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002964009962640099,
+      "loss": 0.299,
+      "step": 2293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002963885429638854,
+      "loss": 0.4448,
+      "step": 2294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029637608966376084,
+      "loss": 0.326,
+      "step": 2295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029636363636363633,
+      "loss": 0.251,
+      "step": 2296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002963511830635118,
+      "loss": 0.2787,
+      "step": 2297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002963387297633873,
+      "loss": 0.5279,
+      "step": 2298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029632627646326273,
+      "loss": 0.4464,
+      "step": 2299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002963138231631382,
+      "loss": 0.5911,
+      "step": 2300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029630136986301365,
+      "loss": 0.315,
+      "step": 2301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029628891656288914,
+      "loss": 0.8973,
+      "step": 2302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002962764632627646,
+      "loss": 0.2625,
+      "step": 2303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029626400996264006,
+      "loss": 0.3888,
+      "step": 2304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029625155666251554,
+      "loss": 0.371,
+      "step": 2305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000296239103362391,
+      "loss": 0.5009,
+      "step": 2306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029622665006226646,
+      "loss": 0.4422,
+      "step": 2307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029621419676214195,
+      "loss": 0.3519,
+      "step": 2308
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029620174346201743,
+      "loss": 0.2885,
+      "step": 2309
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029618929016189286,
+      "loss": 0.3213,
+      "step": 2310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029617683686176835,
+      "loss": 0.2337,
+      "step": 2311
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002961643835616438,
+      "loss": 0.2781,
+      "step": 2312
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029615193026151927,
+      "loss": 0.4429,
+      "step": 2313
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029613947696139476,
+      "loss": 0.2867,
+      "step": 2314
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029612702366127024,
+      "loss": 0.2795,
+      "step": 2315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002961145703611457,
+      "loss": 0.2675,
+      "step": 2316
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029610211706102116,
+      "loss": 0.4155,
+      "step": 2317
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002960896637608966,
+      "loss": 0.3913,
+      "step": 2318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002960772104607721,
+      "loss": 0.5305,
+      "step": 2319
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029606475716064756,
+      "loss": 0.3312,
+      "step": 2320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000296052303860523,
+      "loss": 0.3029,
+      "step": 2321
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002960398505603985,
+      "loss": 0.3555,
+      "step": 2322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002960273972602739,
+      "loss": 0.4027,
+      "step": 2323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002960149439601494,
+      "loss": 0.3271,
+      "step": 2324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002960024906600249,
+      "loss": 0.6481,
+      "step": 2325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002959900373599004,
+      "loss": 0.3261,
+      "step": 2326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002959775840597758,
+      "loss": 0.312,
+      "step": 2327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002959651307596513,
+      "loss": 0.3269,
+      "step": 2328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002959526774595267,
+      "loss": 0.2579,
+      "step": 2329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002959402241594022,
+      "loss": 0.2972,
+      "step": 2330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002959277708592777,
+      "loss": 0.4568,
+      "step": 2331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002959153175591532,
+      "loss": 0.4793,
+      "step": 2332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002959028642590286,
+      "loss": 0.3149,
+      "step": 2333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029589041095890405,
+      "loss": 0.2775,
+      "step": 2334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00029587795765877953,
+      "loss": 0.257,
+      "step": 2335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000295865504358655,
+      "loss": 0.2908,
+      "step": 2336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002958530510585305,
+      "loss": 0.3023,
+      "step": 2337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029584059775840594,
+      "loss": 0.7367,
+      "step": 2338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002958281444582814,
+      "loss": 0.3095,
+      "step": 2339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029581569115815686,
+      "loss": 0.3038,
+      "step": 2340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029580323785803234,
+      "loss": 0.2475,
+      "step": 2341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029579078455790783,
+      "loss": 0.2696,
+      "step": 2342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002957783312577833,
+      "loss": 0.4632,
+      "step": 2343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029576587795765875,
+      "loss": 0.6622,
+      "step": 2344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029575342465753423,
+      "loss": 0.2924,
+      "step": 2345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029574097135740967,
+      "loss": 0.2678,
+      "step": 2346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029572851805728515,
+      "loss": 0.2822,
+      "step": 2347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029571606475716064,
+      "loss": 0.2867,
+      "step": 2348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002957036114570361,
+      "loss": 0.3761,
+      "step": 2349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029569115815691156,
+      "loss": 0.623,
+      "step": 2350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000295678704856787,
+      "loss": 0.2328,
+      "step": 2351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002956662515566625,
+      "loss": 0.4623,
+      "step": 2352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029565379825653796,
+      "loss": 0.2555,
+      "step": 2353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029564134495641345,
+      "loss": 0.2338,
+      "step": 2354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002956288916562889,
+      "loss": 0.385,
+      "step": 2355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029561643835616437,
+      "loss": 0.5894,
+      "step": 2356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002956039850560398,
+      "loss": 0.5543,
+      "step": 2357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002955915317559153,
+      "loss": 0.5737,
+      "step": 2358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029557907845579077,
+      "loss": 0.2791,
+      "step": 2359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00029556662515566626,
+      "loss": 0.3757,
+      "step": 2360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002955541718555417,
+      "loss": 0.3999,
+      "step": 2361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002955417185554172,
+      "loss": 0.3249,
+      "step": 2362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002955292652552926,
+      "loss": 0.7331,
+      "step": 2363
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002955168119551681,
+      "loss": 0.2686,
+      "step": 2364
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002955043586550436,
+      "loss": 0.2991,
+      "step": 2365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000295491905354919,
+      "loss": 0.3794,
+      "step": 2366
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002954794520547945,
+      "loss": 0.2996,
+      "step": 2367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029546699875466993,
+      "loss": 0.4584,
+      "step": 2368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002954545454545454,
+      "loss": 0.4399,
+      "step": 2369
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002954420921544209,
+      "loss": 0.3927,
+      "step": 2370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002954296388542964,
+      "loss": 0.2695,
+      "step": 2371
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002954171855541718,
+      "loss": 0.2888,
+      "step": 2372
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002954047322540473,
+      "loss": 0.2653,
+      "step": 2373
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029539227895392274,
+      "loss": 0.2927,
+      "step": 2374
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002953798256537982,
+      "loss": 0.5758,
+      "step": 2375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002953673723536737,
+      "loss": 0.5215,
+      "step": 2376
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002953549190535492,
+      "loss": 0.3807,
+      "step": 2377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029534246575342463,
+      "loss": 0.2779,
+      "step": 2378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002953300124533001,
+      "loss": 0.3263,
+      "step": 2379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029531755915317555,
+      "loss": 0.2826,
+      "step": 2380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029530510585305103,
+      "loss": 0.341,
+      "step": 2381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002952926525529265,
+      "loss": 0.5482,
+      "step": 2382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029528019925280195,
+      "loss": 0.3837,
+      "step": 2383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029526774595267744,
+      "loss": 0.244,
+      "step": 2384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029525529265255287,
+      "loss": 0.2937,
+      "step": 2385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029524283935242836,
+      "loss": 0.2616,
+      "step": 2386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029523038605230384,
+      "loss": 0.4099,
+      "step": 2387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029521793275217933,
+      "loss": 0.529,
+      "step": 2388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029520547945205476,
+      "loss": 0.3391,
+      "step": 2389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029519302615193025,
+      "loss": 0.3846,
+      "step": 2390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002951805728518057,
+      "loss": 0.2955,
+      "step": 2391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029516811955168117,
+      "loss": 0.2529,
+      "step": 2392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029515566625155665,
+      "loss": 0.4005,
+      "step": 2393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029514321295143214,
+      "loss": 0.5596,
+      "step": 2394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029513075965130757,
+      "loss": 0.2889,
+      "step": 2395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029511830635118306,
+      "loss": 0.3841,
+      "step": 2396
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002951058530510585,
+      "loss": 0.2999,
+      "step": 2397
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000295093399750934,
+      "loss": 0.4432,
+      "step": 2398
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029508094645080946,
+      "loss": 0.3422,
+      "step": 2399
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002950684931506849,
+      "loss": 0.6309,
+      "step": 2400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002950560398505604,
+      "loss": 0.2813,
+      "step": 2401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002950435865504358,
+      "loss": 0.4887,
+      "step": 2402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002950311332503113,
+      "loss": 0.3293,
+      "step": 2403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002950186799501868,
+      "loss": 0.3051,
+      "step": 2404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029500622665006227,
+      "loss": 0.4967,
+      "step": 2405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002949937733499377,
+      "loss": 0.3775,
+      "step": 2406
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002949813200498132,
+      "loss": 0.3994,
+      "step": 2407
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002949688667496886,
+      "loss": 0.2759,
+      "step": 2408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002949564134495641,
+      "loss": 0.2582,
+      "step": 2409
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002949439601494396,
+      "loss": 0.333,
+      "step": 2410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002949315068493151,
+      "loss": 0.3386,
+      "step": 2411
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002949190535491905,
+      "loss": 0.3457,
+      "step": 2412
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00029490660024906595,
+      "loss": 0.4483,
+      "step": 2413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029489414694894143,
+      "loss": 0.2524,
+      "step": 2414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002948816936488169,
+      "loss": 0.3749,
+      "step": 2415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002948692403486924,
+      "loss": 0.3359,
+      "step": 2416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029485678704856784,
+      "loss": 0.2344,
+      "step": 2417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002948443337484433,
+      "loss": 0.3207,
+      "step": 2418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029483188044831875,
+      "loss": 0.6533,
+      "step": 2419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029481942714819424,
+      "loss": 0.2546,
+      "step": 2420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002948069738480697,
+      "loss": 0.2541,
+      "step": 2421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002947945205479452,
+      "loss": 0.3443,
+      "step": 2422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029478206724782065,
+      "loss": 0.3742,
+      "step": 2423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029476961394769613,
+      "loss": 0.3036,
+      "step": 2424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029475716064757156,
+      "loss": 0.5372,
+      "step": 2425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029474470734744705,
+      "loss": 0.3016,
+      "step": 2426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029473225404732254,
+      "loss": 0.3009,
+      "step": 2427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000294719800747198,
+      "loss": 0.2955,
+      "step": 2428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029470734744707345,
+      "loss": 0.2577,
+      "step": 2429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002946948941469489,
+      "loss": 0.3189,
+      "step": 2430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029468244084682437,
+      "loss": 0.4263,
+      "step": 2431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029466998754669986,
+      "loss": 0.4255,
+      "step": 2432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029465753424657535,
+      "loss": 0.4535,
+      "step": 2433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002946450809464508,
+      "loss": 0.3029,
+      "step": 2434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029463262764632626,
+      "loss": 0.3679,
+      "step": 2435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002946201743462017,
+      "loss": 0.3588,
+      "step": 2436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002946077210460772,
+      "loss": 0.3837,
+      "step": 2437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029459526774595267,
+      "loss": 0.5648,
+      "step": 2438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00029458281444582815,
+      "loss": 0.3195,
+      "step": 2439
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002945703611457036,
+      "loss": 0.2923,
+      "step": 2440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029455790784557907,
+      "loss": 0.2656,
+      "step": 2441
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002945454545454545,
+      "loss": 0.2573,
+      "step": 2442
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029453300124533,
+      "loss": 0.3218,
+      "step": 2443
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002945205479452055,
+      "loss": 0.4399,
+      "step": 2444
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002945080946450809,
+      "loss": 0.2884,
+      "step": 2445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002944956413449564,
+      "loss": 0.3559,
+      "step": 2446
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029448318804483183,
+      "loss": 0.3295,
+      "step": 2447
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002944707347447073,
+      "loss": 0.2363,
+      "step": 2448
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002944582814445828,
+      "loss": 0.2593,
+      "step": 2449
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002944458281444583,
+      "loss": 0.5945,
+      "step": 2450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002944333748443337,
+      "loss": 0.2867,
+      "step": 2451
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002944209215442092,
+      "loss": 0.2667,
+      "step": 2452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029440846824408464,
+      "loss": 0.3993,
+      "step": 2453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002943960149439601,
+      "loss": 0.2866,
+      "step": 2454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002943835616438356,
+      "loss": 0.3229,
+      "step": 2455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002943711083437111,
+      "loss": 0.3248,
+      "step": 2456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029435865504358653,
+      "loss": 0.4602,
+      "step": 2457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000294346201743462,
+      "loss": 0.3491,
+      "step": 2458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029433374844333745,
+      "loss": 0.4026,
+      "step": 2459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029432129514321293,
+      "loss": 0.5319,
+      "step": 2460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002943088418430884,
+      "loss": 0.2637,
+      "step": 2461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029429638854296385,
+      "loss": 0.3243,
+      "step": 2462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029428393524283934,
+      "loss": 0.5372,
+      "step": 2463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029427148194271477,
+      "loss": 0.2892,
+      "step": 2464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029425902864259026,
+      "loss": 0.4573,
+      "step": 2465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029424657534246574,
+      "loss": 0.2338,
+      "step": 2466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029423412204234123,
+      "loss": 0.2034,
+      "step": 2467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029422166874221666,
+      "loss": 0.3444,
+      "step": 2468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029420921544209215,
+      "loss": 0.579,
+      "step": 2469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002941967621419676,
+      "loss": 0.3473,
+      "step": 2470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029418430884184306,
+      "loss": 0.3458,
+      "step": 2471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029417185554171855,
+      "loss": 0.3435,
+      "step": 2472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029415940224159404,
+      "loss": 0.2572,
+      "step": 2473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029414694894146947,
+      "loss": 0.2902,
+      "step": 2474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029413449564134496,
+      "loss": 0.5619,
+      "step": 2475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002941220423412204,
+      "loss": 0.3038,
+      "step": 2476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002941095890410959,
+      "loss": 0.476,
+      "step": 2477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029409713574097136,
+      "loss": 0.2921,
+      "step": 2478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002940846824408468,
+      "loss": 0.3712,
+      "step": 2479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002940722291407223,
+      "loss": 0.2854,
+      "step": 2480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002940597758405977,
+      "loss": 0.5911,
+      "step": 2481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002940473225404732,
+      "loss": 0.5396,
+      "step": 2482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002940348692403487,
+      "loss": 0.3556,
+      "step": 2483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029402241594022417,
+      "loss": 0.2454,
+      "step": 2484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002940099626400996,
+      "loss": 0.2729,
+      "step": 2485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002939975093399751,
+      "loss": 0.3905,
+      "step": 2486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002939850560398505,
+      "loss": 0.398,
+      "step": 2487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000293972602739726,
+      "loss": 0.505,
+      "step": 2488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002939601494396015,
+      "loss": 0.2469,
+      "step": 2489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002939476961394769,
+      "loss": 0.326,
+      "step": 2490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002939352428393524,
+      "loss": 0.2903,
+      "step": 2491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029392278953922784,
+      "loss": 0.2627,
+      "step": 2492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029391033623910333,
+      "loss": 0.3589,
+      "step": 2493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002938978829389788,
+      "loss": 0.3892,
+      "step": 2494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002938854296388543,
+      "loss": 0.3623,
+      "step": 2495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029387297633872973,
+      "loss": 0.2839,
+      "step": 2496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002938605230386052,
+      "loss": 0.264,
+      "step": 2497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029384806973848065,
+      "loss": 0.399,
+      "step": 2498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029383561643835614,
+      "loss": 0.3119,
+      "step": 2499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002938231631382316,
+      "loss": 0.5787,
+      "step": 2500
+    },
+    {
+      "epoch": 0.96,
+      "eval_cer": 0.11674112804104805,
+      "eval_loss": 0.42418527603149414,
+      "eval_runtime": 46.9126,
+      "eval_samples_per_second": 37.367,
+      "eval_steps_per_second": 4.69,
+      "eval_wer": 0.21480291442277233,
+      "step": 2500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029381070983810706,
+      "loss": 0.2941,
+      "step": 2501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029379825653798254,
+      "loss": 0.2379,
+      "step": 2502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029378580323785803,
+      "loss": 0.3025,
+      "step": 2503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029377334993773346,
+      "loss": 0.4221,
+      "step": 2504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029376089663760895,
+      "loss": 0.2327,
+      "step": 2505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029374844333748443,
+      "loss": 0.4664,
+      "step": 2506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029373599003735987,
+      "loss": 0.6414,
+      "step": 2507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029372353673723535,
+      "loss": 0.2774,
+      "step": 2508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002937110834371108,
+      "loss": 0.2081,
+      "step": 2509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029369863013698627,
+      "loss": 0.2413,
+      "step": 2510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029368617683686176,
+      "loss": 0.2383,
+      "step": 2511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002936737235367372,
+      "loss": 0.4722,
+      "step": 2512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002936612702366127,
+      "loss": 0.4924,
+      "step": 2513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029364881693648816,
+      "loss": 0.2689,
+      "step": 2514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002936363636363636,
+      "loss": 0.2219,
+      "step": 2515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002936239103362391,
+      "loss": 0.2472,
+      "step": 2516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00029361145703611457,
+      "loss": 0.2864,
+      "step": 2517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029359900373599,
+      "loss": 0.3354,
+      "step": 2518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002935865504358655,
+      "loss": 0.6437,
+      "step": 2519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029357409713574097,
+      "loss": 0.3433,
+      "step": 2520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002935616438356164,
+      "loss": 0.397,
+      "step": 2521
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002935491905354919,
+      "loss": 0.2589,
+      "step": 2522
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002935367372353673,
+      "loss": 0.3011,
+      "step": 2523
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002935242839352428,
+      "loss": 0.3785,
+      "step": 2524
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002935118306351183,
+      "loss": 0.4796,
+      "step": 2525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002934993773349937,
+      "loss": 0.3232,
+      "step": 2526
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002934869240348692,
+      "loss": 0.2757,
+      "step": 2527
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002934744707347447,
+      "loss": 0.2319,
+      "step": 2528
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029346201743462013,
+      "loss": 0.3693,
+      "step": 2529
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002934495641344956,
+      "loss": 0.2807,
+      "step": 2530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002934371108343711,
+      "loss": 0.2738,
+      "step": 2531
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029342465753424653,
+      "loss": 0.6078,
+      "step": 2532
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000293412204234122,
+      "loss": 0.3373,
+      "step": 2533
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029339975093399745,
+      "loss": 0.2415,
+      "step": 2534
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029338729763387294,
+      "loss": 0.2279,
+      "step": 2535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002933748443337484,
+      "loss": 0.2632,
+      "step": 2536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002933623910336239,
+      "loss": 0.3013,
+      "step": 2537
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029334993773349934,
+      "loss": 0.6871,
+      "step": 2538
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002933374844333748,
+      "loss": 0.3131,
+      "step": 2539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029332503113325026,
+      "loss": 0.3164,
+      "step": 2540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029331257783312575,
+      "loss": 0.2594,
+      "step": 2541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029330012453300123,
+      "loss": 0.361,
+      "step": 2542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00029328767123287667,
+      "loss": 0.424,
+      "step": 2543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029327521793275215,
+      "loss": 0.5266,
+      "step": 2544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002932627646326276,
+      "loss": 0.2525,
+      "step": 2545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029325031133250307,
+      "loss": 0.3309,
+      "step": 2546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029323785803237856,
+      "loss": 0.2932,
+      "step": 2547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029322540473225404,
+      "loss": 0.2394,
+      "step": 2548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002932129514321295,
+      "loss": 0.2603,
+      "step": 2549
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029320049813200496,
+      "loss": 0.5567,
+      "step": 2550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002931880448318804,
+      "loss": 0.4686,
+      "step": 2551
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002931755915317559,
+      "loss": 0.3032,
+      "step": 2552
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029316313823163137,
+      "loss": 0.2747,
+      "step": 2553
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002931506849315068,
+      "loss": 0.3429,
+      "step": 2554
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002931382316313823,
+      "loss": 0.3133,
+      "step": 2555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002931257783312577,
+      "loss": 0.5677,
+      "step": 2556
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002931133250311332,
+      "loss": 0.574,
+      "step": 2557
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002931008717310087,
+      "loss": 0.3037,
+      "step": 2558
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002930884184308842,
+      "loss": 0.2825,
+      "step": 2559
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002930759651307596,
+      "loss": 0.2868,
+      "step": 2560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002930635118306351,
+      "loss": 0.2639,
+      "step": 2561
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002930510585305105,
+      "loss": 0.3135,
+      "step": 2562
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000293038605230386,
+      "loss": 0.5184,
+      "step": 2563
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002930261519302615,
+      "loss": 0.3641,
+      "step": 2564
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000293013698630137,
+      "loss": 0.2893,
+      "step": 2565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002930012453300124,
+      "loss": 0.2751,
+      "step": 2566
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002929887920298879,
+      "loss": 0.2937,
+      "step": 2567
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029297633872976334,
+      "loss": 0.3234,
+      "step": 2568
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002929638854296388,
+      "loss": 0.5191,
+      "step": 2569
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002929514321295143,
+      "loss": 0.3311,
+      "step": 2570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029293897882938974,
+      "loss": 0.4253,
+      "step": 2571
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002929265255292652,
+      "loss": 0.2977,
+      "step": 2572
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029291407222914066,
+      "loss": 0.2555,
+      "step": 2573
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029290161892901615,
+      "loss": 0.3395,
+      "step": 2574
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029288916562889163,
+      "loss": 0.5582,
+      "step": 2575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002928767123287671,
+      "loss": 0.3192,
+      "step": 2576
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029286425902864255,
+      "loss": 0.2501,
+      "step": 2577
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029285180572851804,
+      "loss": 0.2334,
+      "step": 2578
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029283935242839347,
+      "loss": 0.2037,
+      "step": 2579
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029282689912826895,
+      "loss": 0.3028,
+      "step": 2580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029281444582814444,
+      "loss": 0.4496,
+      "step": 2581
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002928019925280199,
+      "loss": 0.4051,
+      "step": 2582
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029278953922789536,
+      "loss": 0.2392,
+      "step": 2583
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029277708592777085,
+      "loss": 0.3269,
+      "step": 2584
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002927646326276463,
+      "loss": 0.3151,
+      "step": 2585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029275217932752176,
+      "loss": 0.4156,
+      "step": 2586
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029273972602739725,
+      "loss": 0.5748,
+      "step": 2587
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002927272727272727,
+      "loss": 0.4421,
+      "step": 2588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029271481942714817,
+      "loss": 0.2806,
+      "step": 2589
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002927023661270236,
+      "loss": 0.3658,
+      "step": 2590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002926899128268991,
+      "loss": 0.403,
+      "step": 2591
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029267745952677457,
+      "loss": 0.4358,
+      "step": 2592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00029266500622665006,
+      "loss": 0.4379,
+      "step": 2593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002926525529265255,
+      "loss": 0.799,
+      "step": 2594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000292640099626401,
+      "loss": 0.252,
+      "step": 2595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002926276463262764,
+      "loss": 0.221,
+      "step": 2596
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002926151930261519,
+      "loss": 0.2811,
+      "step": 2597
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002926027397260274,
+      "loss": 0.4511,
+      "step": 2598
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029259028642590287,
+      "loss": 0.806,
+      "step": 2599
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002925778331257783,
+      "loss": 0.5656,
+      "step": 2600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029256537982565373,
+      "loss": 0.4193,
+      "step": 2601
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002925529265255292,
+      "loss": 0.3185,
+      "step": 2602
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002925404732254047,
+      "loss": 0.255,
+      "step": 2603
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002925280199252802,
+      "loss": 0.2691,
+      "step": 2604
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002925155666251556,
+      "loss": 0.2671,
+      "step": 2605
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002925031133250311,
+      "loss": 0.4291,
+      "step": 2606
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029249066002490654,
+      "loss": 0.4759,
+      "step": 2607
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029247820672478203,
+      "loss": 0.3172,
+      "step": 2608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002924657534246575,
+      "loss": 0.5419,
+      "step": 2609
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000292453300124533,
+      "loss": 0.3188,
+      "step": 2610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029244084682440843,
+      "loss": 0.2544,
+      "step": 2611
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002924283935242839,
+      "loss": 0.4865,
+      "step": 2612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029241594022415935,
+      "loss": 0.4698,
+      "step": 2613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029240348692403484,
+      "loss": 0.4871,
+      "step": 2614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002923910336239103,
+      "loss": 0.4833,
+      "step": 2615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002923785803237858,
+      "loss": 0.6255,
+      "step": 2616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029236612702366124,
+      "loss": 0.3015,
+      "step": 2617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002923536737235367,
+      "loss": 0.3628,
+      "step": 2618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029234122042341216,
+      "loss": 0.25,
+      "step": 2619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029232876712328765,
+      "loss": 0.3063,
+      "step": 2620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029231631382316313,
+      "loss": 0.2202,
+      "step": 2621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029230386052303856,
+      "loss": 0.7092,
+      "step": 2622
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029229140722291405,
+      "loss": 0.3077,
+      "step": 2623
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002922789539227895,
+      "loss": 0.448,
+      "step": 2624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029226650062266497,
+      "loss": 0.3004,
+      "step": 2625
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029225404732254046,
+      "loss": 0.3074,
+      "step": 2626
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029224159402241594,
+      "loss": 0.424,
+      "step": 2627
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002922291407222914,
+      "loss": 0.5248,
+      "step": 2628
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029221668742216686,
+      "loss": 0.4346,
+      "step": 2629
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002922042341220423,
+      "loss": 0.326,
+      "step": 2630
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002921917808219178,
+      "loss": 0.2916,
+      "step": 2631
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029217932752179326,
+      "loss": 0.2507,
+      "step": 2632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002921668742216687,
+      "loss": 0.3582,
+      "step": 2633
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002921544209215442,
+      "loss": 0.4635,
+      "step": 2634
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002921419676214196,
+      "loss": 0.3765,
+      "step": 2635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002921295143212951,
+      "loss": 0.2816,
+      "step": 2636
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002921170610211706,
+      "loss": 0.3224,
+      "step": 2637
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002921046077210461,
+      "loss": 0.2551,
+      "step": 2638
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002920921544209215,
+      "loss": 0.3256,
+      "step": 2639
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000292079701120797,
+      "loss": 0.2911,
+      "step": 2640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002920672478206724,
+      "loss": 0.3167,
+      "step": 2641
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002920547945205479,
+      "loss": 0.284,
+      "step": 2642
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002920423412204234,
+      "loss": 0.2412,
+      "step": 2643
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002920298879202989,
+      "loss": 0.202,
+      "step": 2644
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002920174346201743,
+      "loss": 0.3673,
+      "step": 2645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002920049813200498,
+      "loss": 0.3259,
+      "step": 2646
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029199252801992523,
+      "loss": 0.3954,
+      "step": 2647
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002919800747198007,
+      "loss": 0.4535,
+      "step": 2648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002919676214196762,
+      "loss": 0.284,
+      "step": 2649
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029195516811955164,
+      "loss": 0.2984,
+      "step": 2650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002919427148194271,
+      "loss": 0.2815,
+      "step": 2651
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029193026151930256,
+      "loss": 0.2677,
+      "step": 2652
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029191780821917804,
+      "loss": 0.54,
+      "step": 2653
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029190535491905353,
+      "loss": 0.3012,
+      "step": 2654
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000291892901618929,
+      "loss": 0.2932,
+      "step": 2655
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029188044831880445,
+      "loss": 0.3098,
+      "step": 2656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029186799501867993,
+      "loss": 0.2925,
+      "step": 2657
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029185554171855537,
+      "loss": 0.2202,
+      "step": 2658
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029184308841843085,
+      "loss": 0.5133,
+      "step": 2659
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029183063511830634,
+      "loss": 0.2386,
+      "step": 2660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002918181818181818,
+      "loss": 0.3479,
+      "step": 2661
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029180572851805726,
+      "loss": 0.2718,
+      "step": 2662
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029179327521793274,
+      "loss": 0.2623,
+      "step": 2663
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002917808219178082,
+      "loss": 0.3748,
+      "step": 2664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029176836861768366,
+      "loss": 0.5276,
+      "step": 2665
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029175591531755915,
+      "loss": 0.5172,
+      "step": 2666
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002917434620174346,
+      "loss": 0.25,
+      "step": 2667
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029173100871731007,
+      "loss": 0.4695,
+      "step": 2668
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002917185554171855,
+      "loss": 0.2422,
+      "step": 2669
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000291706102117061,
+      "loss": 0.3019,
+      "step": 2670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029169364881693647,
+      "loss": 0.368,
+      "step": 2671
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029168119551681196,
+      "loss": 0.5196,
+      "step": 2672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002916687422166874,
+      "loss": 0.4284,
+      "step": 2673
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002916562889165629,
+      "loss": 0.1844,
+      "step": 2674
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002916438356164383,
+      "loss": 0.224,
+      "step": 2675
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002916313823163138,
+      "loss": 0.2469,
+      "step": 2676
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002916189290161893,
+      "loss": 0.2676,
+      "step": 2677
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029160647571606477,
+      "loss": 0.468,
+      "step": 2678
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002915940224159402,
+      "loss": 0.363,
+      "step": 2679
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029158156911581563,
+      "loss": 0.2271,
+      "step": 2680
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002915691158156911,
+      "loss": 0.3325,
+      "step": 2681
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002915566625155666,
+      "loss": 0.3497,
+      "step": 2682
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002915442092154421,
+      "loss": 0.2313,
+      "step": 2683
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002915317559153175,
+      "loss": 0.4984,
+      "step": 2684
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000291519302615193,
+      "loss": 0.3553,
+      "step": 2685
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029150684931506844,
+      "loss": 0.2476,
+      "step": 2686
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002914943960149439,
+      "loss": 0.2316,
+      "step": 2687
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002914819427148194,
+      "loss": 0.2868,
+      "step": 2688
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002914694894146949,
+      "loss": 0.2321,
+      "step": 2689
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029145703611457033,
+      "loss": 0.4157,
+      "step": 2690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002914445828144458,
+      "loss": 0.9279,
+      "step": 2691
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029143212951432125,
+      "loss": 0.3556,
+      "step": 2692
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029141967621419673,
+      "loss": 0.3926,
+      "step": 2693
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002914072229140722,
+      "loss": 0.3285,
+      "step": 2694
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029139476961394765,
+      "loss": 0.291,
+      "step": 2695
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029138231631382314,
+      "loss": 0.2893,
+      "step": 2696
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029136986301369857,
+      "loss": 0.4024,
+      "step": 2697
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029135740971357406,
+      "loss": 0.3113,
+      "step": 2698
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029134495641344954,
+      "loss": 0.2835,
+      "step": 2699
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00029133250311332503,
+      "loss": 0.2771,
+      "step": 2700
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029132004981320046,
+      "loss": 0.2508,
+      "step": 2701
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029130759651307595,
+      "loss": 0.5779,
+      "step": 2702
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002912951432129514,
+      "loss": 0.4523,
+      "step": 2703
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029128268991282687,
+      "loss": 0.4925,
+      "step": 2704
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029127023661270235,
+      "loss": 0.3728,
+      "step": 2705
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029125778331257784,
+      "loss": 0.2576,
+      "step": 2706
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029124533001245327,
+      "loss": 0.3074,
+      "step": 2707
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029123287671232876,
+      "loss": 0.3232,
+      "step": 2708
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002912204234122042,
+      "loss": 0.707,
+      "step": 2709
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002912079701120797,
+      "loss": 0.5,
+      "step": 2710
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029119551681195516,
+      "loss": 0.3274,
+      "step": 2711
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002911830635118306,
+      "loss": 0.333,
+      "step": 2712
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002911706102117061,
+      "loss": 0.4251,
+      "step": 2713
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002911581569115815,
+      "loss": 0.3417,
+      "step": 2714
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000291145703611457,
+      "loss": 0.3602,
+      "step": 2715
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002911332503113325,
+      "loss": 0.435,
+      "step": 2716
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029112079701120797,
+      "loss": 0.303,
+      "step": 2717
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002911083437110834,
+      "loss": 0.277,
+      "step": 2718
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002910958904109589,
+      "loss": 0.231,
+      "step": 2719
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002910834371108343,
+      "loss": 0.2409,
+      "step": 2720
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002910709838107098,
+      "loss": 0.5582,
+      "step": 2721
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002910585305105853,
+      "loss": 0.4076,
+      "step": 2722
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002910460772104608,
+      "loss": 0.2578,
+      "step": 2723
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002910336239103362,
+      "loss": 0.4793,
+      "step": 2724
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002910211706102117,
+      "loss": 0.3878,
+      "step": 2725
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00029100871731008713,
+      "loss": 0.3119,
+      "step": 2726
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002909962640099626,
+      "loss": 0.3817,
+      "step": 2727
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002909838107098381,
+      "loss": 0.7592,
+      "step": 2728
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029097135740971354,
+      "loss": 0.3566,
+      "step": 2729
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000290958904109589,
+      "loss": 0.2515,
+      "step": 2730
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029094645080946445,
+      "loss": 0.4217,
+      "step": 2731
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029093399750933994,
+      "loss": 0.2443,
+      "step": 2732
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002909215442092154,
+      "loss": 0.3562,
+      "step": 2733
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002909090909090909,
+      "loss": 0.5744,
+      "step": 2734
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029089663760896635,
+      "loss": 0.3241,
+      "step": 2735
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029088418430884183,
+      "loss": 0.2473,
+      "step": 2736
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029087173100871726,
+      "loss": 0.2825,
+      "step": 2737
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029085927770859275,
+      "loss": 0.3698,
+      "step": 2738
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029084682440846824,
+      "loss": 0.2021,
+      "step": 2739
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002908343711083437,
+      "loss": 0.314,
+      "step": 2740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029082191780821915,
+      "loss": 0.7816,
+      "step": 2741
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029080946450809464,
+      "loss": 0.2842,
+      "step": 2742
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029079701120797007,
+      "loss": 0.2563,
+      "step": 2743
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029078455790784556,
+      "loss": 0.3671,
+      "step": 2744
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029077210460772105,
+      "loss": 0.3465,
+      "step": 2745
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002907596513075965,
+      "loss": 0.419,
+      "step": 2746
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029074719800747196,
+      "loss": 0.5611,
+      "step": 2747
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002907347447073474,
+      "loss": 0.4632,
+      "step": 2748
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002907222914072229,
+      "loss": 0.4172,
+      "step": 2749
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029070983810709837,
+      "loss": 0.3414,
+      "step": 2750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029069738480697385,
+      "loss": 0.3183,
+      "step": 2751
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002906849315068493,
+      "loss": 0.3332,
+      "step": 2752
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002906724782067248,
+      "loss": 0.6203,
+      "step": 2753
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002906600249066002,
+      "loss": 0.4136,
+      "step": 2754
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002906475716064757,
+      "loss": 0.2155,
+      "step": 2755
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002906351183063512,
+      "loss": 0.3411,
+      "step": 2756
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029062266500622666,
+      "loss": 0.2388,
+      "step": 2757
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002906102117061021,
+      "loss": 0.2756,
+      "step": 2758
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029059775840597753,
+      "loss": 0.4814,
+      "step": 2759
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000290585305105853,
+      "loss": 0.4526,
+      "step": 2760
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002905728518057285,
+      "loss": 0.346,
+      "step": 2761
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000290560398505604,
+      "loss": 0.2533,
+      "step": 2762
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002905479452054794,
+      "loss": 0.2432,
+      "step": 2763
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002905354919053549,
+      "loss": 0.2358,
+      "step": 2764
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029052303860523034,
+      "loss": 0.4356,
+      "step": 2765
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002905105853051058,
+      "loss": 0.3589,
+      "step": 2766
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002904981320049813,
+      "loss": 0.2817,
+      "step": 2767
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002904856787048568,
+      "loss": 0.3035,
+      "step": 2768
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029047322540473223,
+      "loss": 0.4239,
+      "step": 2769
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002904607721046077,
+      "loss": 0.2634,
+      "step": 2770
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029044831880448315,
+      "loss": 0.1721,
+      "step": 2771
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029043586550435863,
+      "loss": 0.4622,
+      "step": 2772
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002904234122042341,
+      "loss": 0.2857,
+      "step": 2773
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029041095890410955,
+      "loss": 0.3285,
+      "step": 2774
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029039850560398504,
+      "loss": 0.3379,
+      "step": 2775
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029038605230386047,
+      "loss": 0.2568,
+      "step": 2776
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029037359900373596,
+      "loss": 0.2363,
+      "step": 2777
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00029036114570361144,
+      "loss": 0.6495,
+      "step": 2778
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029034869240348693,
+      "loss": 0.3573,
+      "step": 2779
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029033623910336236,
+      "loss": 0.2759,
+      "step": 2780
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029032378580323785,
+      "loss": 0.3015,
+      "step": 2781
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002903113325031133,
+      "loss": 0.2989,
+      "step": 2782
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029029887920298876,
+      "loss": 0.3478,
+      "step": 2783
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029028642590286425,
+      "loss": 0.4749,
+      "step": 2784
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029027397260273974,
+      "loss": 0.2682,
+      "step": 2785
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029026151930261517,
+      "loss": 0.3003,
+      "step": 2786
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029024906600249066,
+      "loss": 0.282,
+      "step": 2787
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002902366127023661,
+      "loss": 0.2628,
+      "step": 2788
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002902241594022416,
+      "loss": 0.4148,
+      "step": 2789
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029021170610211706,
+      "loss": 0.5057,
+      "step": 2790
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002901992528019925,
+      "loss": 0.4953,
+      "step": 2791
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000290186799501868,
+      "loss": 0.3897,
+      "step": 2792
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002901743462017434,
+      "loss": 0.3706,
+      "step": 2793
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002901618929016189,
+      "loss": 0.1939,
+      "step": 2794
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002901494396014944,
+      "loss": 0.3166,
+      "step": 2795
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00029013698630136987,
+      "loss": 0.3336,
+      "step": 2796
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002901245330012453,
+      "loss": 0.466,
+      "step": 2797
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002901120797011208,
+      "loss": 0.2998,
+      "step": 2798
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002900996264009962,
+      "loss": 0.2855,
+      "step": 2799
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002900871731008717,
+      "loss": 0.3864,
+      "step": 2800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002900747198007472,
+      "loss": 0.2282,
+      "step": 2801
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002900622665006227,
+      "loss": 0.2352,
+      "step": 2802
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002900498132004981,
+      "loss": 0.489,
+      "step": 2803
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002900373599003736,
+      "loss": 0.3841,
+      "step": 2804
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029002490660024903,
+      "loss": 0.3083,
+      "step": 2805
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002900124533001245,
+      "loss": 0.2026,
+      "step": 2806
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00029,
+      "loss": 0.2095,
+      "step": 2807
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028998754669987543,
+      "loss": 0.2668,
+      "step": 2808
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002899750933997509,
+      "loss": 0.5031,
+      "step": 2809
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028996264009962635,
+      "loss": 0.2855,
+      "step": 2810
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028995018679950184,
+      "loss": 0.3059,
+      "step": 2811
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002899377334993773,
+      "loss": 0.417,
+      "step": 2812
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002899252801992528,
+      "loss": 0.2715,
+      "step": 2813
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028991282689912824,
+      "loss": 0.2409,
+      "step": 2814
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028990037359900373,
+      "loss": 0.3925,
+      "step": 2815
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028988792029887916,
+      "loss": 0.3704,
+      "step": 2816
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028987546699875465,
+      "loss": 0.258,
+      "step": 2817
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028986301369863013,
+      "loss": 0.3451,
+      "step": 2818
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002898505603985056,
+      "loss": 0.2591,
+      "step": 2819
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028983810709838105,
+      "loss": 0.2761,
+      "step": 2820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002898256537982565,
+      "loss": 0.3025,
+      "step": 2821
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028981320049813197,
+      "loss": 0.4527,
+      "step": 2822
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028980074719800746,
+      "loss": 0.2469,
+      "step": 2823
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028978829389788294,
+      "loss": 0.4097,
+      "step": 2824
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002897758405977584,
+      "loss": 0.249,
+      "step": 2825
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028976338729763386,
+      "loss": 0.2549,
+      "step": 2826
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002897509339975093,
+      "loss": 0.2485,
+      "step": 2827
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002897384806973848,
+      "loss": 0.3604,
+      "step": 2828
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028972602739726027,
+      "loss": 0.2946,
+      "step": 2829
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00028971357409713575,
+      "loss": 0.285,
+      "step": 2830
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002897011207970112,
+      "loss": 0.4141,
+      "step": 2831
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00028968866749688667,
+      "loss": 0.3459,
+      "step": 2832
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002896762141967621,
+      "loss": 0.3055,
+      "step": 2833
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002896637608966376,
+      "loss": 0.4074,
+      "step": 2834
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002896513075965131,
+      "loss": 0.2592,
+      "step": 2835
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002896388542963885,
+      "loss": 0.3059,
+      "step": 2836
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000289626400996264,
+      "loss": 0.2154,
+      "step": 2837
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002896139476961394,
+      "loss": 0.2532,
+      "step": 2838
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002896014943960149,
+      "loss": 0.271,
+      "step": 2839
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002895890410958904,
+      "loss": 0.2688,
+      "step": 2840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002895765877957659,
+      "loss": 0.5221,
+      "step": 2841
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002895641344956413,
+      "loss": 0.2794,
+      "step": 2842
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002895516811955168,
+      "loss": 0.22,
+      "step": 2843
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00028953922789539223,
+      "loss": 0.3317,
+      "step": 2844
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002895267745952677,
+      "loss": 0.2234,
+      "step": 2845
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002895143212951432,
+      "loss": 0.2499,
+      "step": 2846
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00028950186799501864,
+      "loss": 0.4196,
+      "step": 2847
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002894894146948941,
+      "loss": 0.3546,
+      "step": 2848
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002894769613947696,
+      "loss": 0.234,
+      "step": 2849
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00028946450809464504,
+      "loss": 0.2257,
+      "step": 2850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00028945205479452053,
+      "loss": 0.3484,
+      "step": 2851
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000289439601494396,
+      "loss": 0.2866,
+      "step": 2852
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00028942714819427145,
+      "loss": 0.6866,
+      "step": 2853
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00028941469489414693,
+      "loss": 0.249,
+      "step": 2854
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00028940224159402237,
+      "loss": 0.2924,
+      "step": 2855
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00028938978829389785,
+      "loss": 0.2538,
+      "step": 2856
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00028937733499377334,
+      "loss": 0.4046,
+      "step": 2857
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00028936488169364877,
+      "loss": 0.2024,
+      "step": 2858
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00028935242839352426,
+      "loss": 0.4751,
+      "step": 2859
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00028933997509339974,
+      "loss": 0.3027,
+      "step": 2860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002893275217932752,
+      "loss": 0.4347,
+      "step": 2861
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00028931506849315066,
+      "loss": 0.353,
+      "step": 2862
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002893026151930261,
+      "loss": 0.2822,
+      "step": 2863
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002892901618929016,
+      "loss": 0.2657,
+      "step": 2864
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00028927770859277707,
+      "loss": 0.3755,
+      "step": 2865
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00028926525529265255,
+      "loss": 0.4501,
+      "step": 2866
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000289252801992528,
+      "loss": 0.2439,
+      "step": 2867
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00028924034869240347,
+      "loss": 0.2813,
+      "step": 2868
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002892278953922789,
+      "loss": 0.2415,
+      "step": 2869
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002892154420921544,
+      "loss": 0.2801,
+      "step": 2870
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002892029887920299,
+      "loss": 0.3688,
+      "step": 2871
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002891905354919053,
+      "loss": 0.5275,
+      "step": 2872
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002891780821917808,
+      "loss": 0.2929,
+      "step": 2873
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002891656288916562,
+      "loss": 0.3377,
+      "step": 2874
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002891531755915317,
+      "loss": 0.2274,
+      "step": 2875
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002891407222914072,
+      "loss": 0.2555,
+      "step": 2876
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002891282689912827,
+      "loss": 0.2687,
+      "step": 2877
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002891158156911581,
+      "loss": 0.4879,
+      "step": 2878
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002891033623910336,
+      "loss": 0.3229,
+      "step": 2879
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00028909090909090904,
+      "loss": 0.2787,
+      "step": 2880
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002890784557907845,
+      "loss": 0.2715,
+      "step": 2881
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00028906600249066,
+      "loss": 0.199,
+      "step": 2882
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002890535491905355,
+      "loss": 0.1855,
+      "step": 2883
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002890410958904109,
+      "loss": 0.5545,
+      "step": 2884
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028902864259028636,
+      "loss": 0.3582,
+      "step": 2885
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028901618929016185,
+      "loss": 0.2635,
+      "step": 2886
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028900373599003733,
+      "loss": 0.3738,
+      "step": 2887
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002889912826899128,
+      "loss": 0.2793,
+      "step": 2888
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028897882938978825,
+      "loss": 0.2097,
+      "step": 2889
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028896637608966374,
+      "loss": 0.4967,
+      "step": 2890
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028895392278953917,
+      "loss": 0.4133,
+      "step": 2891
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028894146948941465,
+      "loss": 0.4267,
+      "step": 2892
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028892901618929014,
+      "loss": 0.3634,
+      "step": 2893
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002889165628891656,
+      "loss": 0.2664,
+      "step": 2894
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028890410958904106,
+      "loss": 0.3455,
+      "step": 2895
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028889165628891655,
+      "loss": 0.3396,
+      "step": 2896
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000288879202988792,
+      "loss": 0.4686,
+      "step": 2897
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028886674968866746,
+      "loss": 0.5165,
+      "step": 2898
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028885429638854295,
+      "loss": 0.2731,
+      "step": 2899
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002888418430884184,
+      "loss": 0.2432,
+      "step": 2900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028882938978829387,
+      "loss": 0.3647,
+      "step": 2901
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002888169364881693,
+      "loss": 0.4611,
+      "step": 2902
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002888044831880448,
+      "loss": 0.4043,
+      "step": 2903
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002887920298879203,
+      "loss": 0.2798,
+      "step": 2904
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028877957658779576,
+      "loss": 0.3637,
+      "step": 2905
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002887671232876712,
+      "loss": 0.2422,
+      "step": 2906
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002887546699875467,
+      "loss": 0.2671,
+      "step": 2907
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002887422166874221,
+      "loss": 0.3246,
+      "step": 2908
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002887297633872976,
+      "loss": 0.5283,
+      "step": 2909
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002887173100871731,
+      "loss": 0.3952,
+      "step": 2910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028870485678704857,
+      "loss": 0.3309,
+      "step": 2911
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000288692403486924,
+      "loss": 0.2257,
+      "step": 2912
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002886799501867995,
+      "loss": 0.3832,
+      "step": 2913
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002886674968866749,
+      "loss": 0.2297,
+      "step": 2914
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002886550435865504,
+      "loss": 0.3914,
+      "step": 2915
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002886425902864259,
+      "loss": 0.4245,
+      "step": 2916
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002886301369863013,
+      "loss": 0.2385,
+      "step": 2917
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002886176836861768,
+      "loss": 0.2241,
+      "step": 2918
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028860523038605224,
+      "loss": 0.2434,
+      "step": 2919
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028859277708592773,
+      "loss": 0.3318,
+      "step": 2920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002885803237858032,
+      "loss": 0.326,
+      "step": 2921
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002885678704856787,
+      "loss": 0.5729,
+      "step": 2922
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028855541718555413,
+      "loss": 0.278,
+      "step": 2923
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002885429638854296,
+      "loss": 0.2907,
+      "step": 2924
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028853051058530505,
+      "loss": 0.2711,
+      "step": 2925
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028851805728518054,
+      "loss": 0.2332,
+      "step": 2926
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000288505603985056,
+      "loss": 0.3422,
+      "step": 2927
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002884931506849315,
+      "loss": 0.497,
+      "step": 2928
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028848069738480694,
+      "loss": 0.257,
+      "step": 2929
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028846824408468243,
+      "loss": 0.3242,
+      "step": 2930
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028845579078455786,
+      "loss": 0.2472,
+      "step": 2931
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028844333748443335,
+      "loss": 0.319,
+      "step": 2932
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028843088418430883,
+      "loss": 0.3372,
+      "step": 2933
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028841843088418426,
+      "loss": 0.5791,
+      "step": 2934
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00028840597758405975,
+      "loss": 0.341,
+      "step": 2935
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002883935242839352,
+      "loss": 0.2234,
+      "step": 2936
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028838107098381067,
+      "loss": 0.2365,
+      "step": 2937
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028836861768368616,
+      "loss": 0.2763,
+      "step": 2938
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028835616438356164,
+      "loss": 0.3579,
+      "step": 2939
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002883437110834371,
+      "loss": 0.4248,
+      "step": 2940
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028833125778331256,
+      "loss": 0.4901,
+      "step": 2941
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000288318804483188,
+      "loss": 0.2202,
+      "step": 2942
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002883063511830635,
+      "loss": 0.2511,
+      "step": 2943
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028829389788293896,
+      "loss": 0.2464,
+      "step": 2944
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028828144458281445,
+      "loss": 0.2698,
+      "step": 2945
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002882689912826899,
+      "loss": 0.4506,
+      "step": 2946
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002882565379825653,
+      "loss": 0.6463,
+      "step": 2947
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002882440846824408,
+      "loss": 0.3105,
+      "step": 2948
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002882316313823163,
+      "loss": 0.2793,
+      "step": 2949
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002882191780821918,
+      "loss": 0.2309,
+      "step": 2950
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002882067247820672,
+      "loss": 0.2243,
+      "step": 2951
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002881942714819427,
+      "loss": 0.3505,
+      "step": 2952
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002881818181818181,
+      "loss": 0.6786,
+      "step": 2953
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002881693648816936,
+      "loss": 0.6207,
+      "step": 2954
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002881569115815691,
+      "loss": 0.4597,
+      "step": 2955
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002881444582814446,
+      "loss": 0.2732,
+      "step": 2956
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028813200498132,
+      "loss": 0.2839,
+      "step": 2957
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002881195516811955,
+      "loss": 0.2801,
+      "step": 2958
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00028810709838107093,
+      "loss": 0.56,
+      "step": 2959
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002880946450809464,
+      "loss": 0.2808,
+      "step": 2960
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002880821917808219,
+      "loss": 0.2517,
+      "step": 2961
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028806973848069734,
+      "loss": 0.2858,
+      "step": 2962
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002880572851805728,
+      "loss": 0.3019,
+      "step": 2963
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028804483188044826,
+      "loss": 0.262,
+      "step": 2964
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028803237858032374,
+      "loss": 0.3132,
+      "step": 2965
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028801992528019923,
+      "loss": 0.5048,
+      "step": 2966
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002880074719800747,
+      "loss": 0.2783,
+      "step": 2967
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028799501867995015,
+      "loss": 0.2641,
+      "step": 2968
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028798256537982563,
+      "loss": 0.2736,
+      "step": 2969
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028797011207970107,
+      "loss": 0.3072,
+      "step": 2970
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028795765877957655,
+      "loss": 0.3042,
+      "step": 2971
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028794520547945204,
+      "loss": 0.4421,
+      "step": 2972
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002879327521793275,
+      "loss": 0.3102,
+      "step": 2973
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028792029887920296,
+      "loss": 0.2941,
+      "step": 2974
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028790784557907844,
+      "loss": 0.6062,
+      "step": 2975
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002878953922789539,
+      "loss": 0.2441,
+      "step": 2976
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028788293897882936,
+      "loss": 0.3261,
+      "step": 2977
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028787048567870485,
+      "loss": 0.4166,
+      "step": 2978
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002878580323785803,
+      "loss": 0.5085,
+      "step": 2979
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028784557907845577,
+      "loss": 0.3035,
+      "step": 2980
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002878331257783312,
+      "loss": 0.2326,
+      "step": 2981
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002878206724782067,
+      "loss": 0.2063,
+      "step": 2982
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028780821917808217,
+      "loss": 0.2871,
+      "step": 2983
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028779576587795766,
+      "loss": 0.5409,
+      "step": 2984
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002877833125778331,
+      "loss": 0.281,
+      "step": 2985
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002877708592777086,
+      "loss": 0.3979,
+      "step": 2986
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000287758405977584,
+      "loss": 0.3092,
+      "step": 2987
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002877459526774595,
+      "loss": 0.2745,
+      "step": 2988
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000287733499377335,
+      "loss": 0.3574,
+      "step": 2989
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028772104607721047,
+      "loss": 0.3528,
+      "step": 2990
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002877085927770859,
+      "loss": 0.4383,
+      "step": 2991
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002876961394769614,
+      "loss": 0.2437,
+      "step": 2992
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002876836861768368,
+      "loss": 0.3501,
+      "step": 2993
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002876712328767123,
+      "loss": 0.2406,
+      "step": 2994
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002876587795765878,
+      "loss": 0.2351,
+      "step": 2995
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002876463262764632,
+      "loss": 0.3795,
+      "step": 2996
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002876338729763387,
+      "loss": 0.4467,
+      "step": 2997
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028762141967621414,
+      "loss": 0.252,
+      "step": 2998
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002876089663760896,
+      "loss": 0.2197,
+      "step": 2999
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002875965130759651,
+      "loss": 0.3465,
+      "step": 3000
+    },
+    {
+      "epoch": 1.15,
+      "eval_cer": 0.11541838187954816,
+      "eval_loss": 0.42104285955429077,
+      "eval_runtime": 47.6468,
+      "eval_samples_per_second": 36.792,
+      "eval_steps_per_second": 4.617,
+      "eval_wer": 0.20869348780377428,
+      "step": 3000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002875840597758406,
+      "loss": 0.2637,
+      "step": 3001
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028757160647571603,
+      "loss": 0.2609,
+      "step": 3002
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002875591531755915,
+      "loss": 0.532,
+      "step": 3003
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028754669987546695,
+      "loss": 0.5963,
+      "step": 3004
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028753424657534243,
+      "loss": 0.2868,
+      "step": 3005
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002875217932752179,
+      "loss": 0.3141,
+      "step": 3006
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002875093399750934,
+      "loss": 0.2126,
+      "step": 3007
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028749688667496884,
+      "loss": 0.3815,
+      "step": 3008
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002874844333748443,
+      "loss": 0.4558,
+      "step": 3009
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028747198007471976,
+      "loss": 0.2974,
+      "step": 3010
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028745952677459524,
+      "loss": 0.4026,
+      "step": 3011
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028744707347447073,
+      "loss": 0.2579,
+      "step": 3012
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028743462017434616,
+      "loss": 0.2845,
+      "step": 3013
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028742216687422165,
+      "loss": 0.2066,
+      "step": 3014
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002874097135740971,
+      "loss": 0.2807,
+      "step": 3015
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028739726027397257,
+      "loss": 0.3325,
+      "step": 3016
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028738480697384805,
+      "loss": 0.3732,
+      "step": 3017
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028737235367372354,
+      "loss": 0.358,
+      "step": 3018
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028735990037359897,
+      "loss": 0.3208,
+      "step": 3019
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028734744707347446,
+      "loss": 0.3243,
+      "step": 3020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002873349937733499,
+      "loss": 0.5508,
+      "step": 3021
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002873225404732254,
+      "loss": 0.418,
+      "step": 3022
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028731008717310086,
+      "loss": 0.3578,
+      "step": 3023
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028729763387297635,
+      "loss": 0.2445,
+      "step": 3024
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002872851805728518,
+      "loss": 0.2877,
+      "step": 3025
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002872727272727272,
+      "loss": 0.2493,
+      "step": 3026
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002872602739726027,
+      "loss": 0.3626,
+      "step": 3027
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002872478206724782,
+      "loss": 0.5855,
+      "step": 3028
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028723536737235367,
+      "loss": 0.2802,
+      "step": 3029
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002872229140722291,
+      "loss": 0.4482,
+      "step": 3030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002872104607721046,
+      "loss": 0.2497,
+      "step": 3031
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028719800747198,
+      "loss": 0.3246,
+      "step": 3032
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002871855541718555,
+      "loss": 0.3018,
+      "step": 3033
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000287173100871731,
+      "loss": 0.6594,
+      "step": 3034
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002871606475716065,
+      "loss": 0.3505,
+      "step": 3035
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002871481942714819,
+      "loss": 0.2193,
+      "step": 3036
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002871357409713574,
+      "loss": 0.3722,
+      "step": 3037
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028712328767123283,
+      "loss": 0.3467,
+      "step": 3038
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002871108343711083,
+      "loss": 0.2976,
+      "step": 3039
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002870983810709838,
+      "loss": 0.4912,
+      "step": 3040
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028708592777085924,
+      "loss": 0.3632,
+      "step": 3041
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002870734744707347,
+      "loss": 0.3321,
+      "step": 3042
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028706102117061015,
+      "loss": 0.2958,
+      "step": 3043
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028704856787048564,
+      "loss": 0.3017,
+      "step": 3044
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002870361145703611,
+      "loss": 0.3044,
+      "step": 3045
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002870236612702366,
+      "loss": 0.3972,
+      "step": 3046
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028701120797011205,
+      "loss": 0.5908,
+      "step": 3047
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028699875466998753,
+      "loss": 0.4136,
+      "step": 3048
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028698630136986296,
+      "loss": 0.2735,
+      "step": 3049
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028697384806973845,
+      "loss": 0.3089,
+      "step": 3050
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028696139476961394,
+      "loss": 0.2984,
+      "step": 3051
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002869489414694894,
+      "loss": 0.2568,
+      "step": 3052
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028693648816936485,
+      "loss": 0.4899,
+      "step": 3053
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028692403486924034,
+      "loss": 0.3094,
+      "step": 3054
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002869115815691158,
+      "loss": 0.2432,
+      "step": 3055
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028689912826899126,
+      "loss": 0.2252,
+      "step": 3056
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028688667496886675,
+      "loss": 0.2296,
+      "step": 3057
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002868742216687422,
+      "loss": 0.2138,
+      "step": 3058
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028686176836861766,
+      "loss": 0.4363,
+      "step": 3059
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002868493150684931,
+      "loss": 0.3534,
+      "step": 3060
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002868368617683686,
+      "loss": 0.2817,
+      "step": 3061
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028682440846824407,
+      "loss": 0.3979,
+      "step": 3062
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00028681195516811955,
+      "loss": 0.3921,
+      "step": 3063
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000286799501867995,
+      "loss": 0.334,
+      "step": 3064
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002867870485678705,
+      "loss": 0.3243,
+      "step": 3065
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002867745952677459,
+      "loss": 0.4275,
+      "step": 3066
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002867621419676214,
+      "loss": 0.5032,
+      "step": 3067
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002867496886674969,
+      "loss": 0.2507,
+      "step": 3068
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028673723536737236,
+      "loss": 0.2754,
+      "step": 3069
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002867247820672478,
+      "loss": 0.2976,
+      "step": 3070
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002867123287671233,
+      "loss": 0.4929,
+      "step": 3071
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002866998754669987,
+      "loss": 0.5233,
+      "step": 3072
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002866874221668742,
+      "loss": 0.2616,
+      "step": 3073
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002866749688667497,
+      "loss": 0.2368,
+      "step": 3074
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002866625155666251,
+      "loss": 0.2079,
+      "step": 3075
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002866500622665006,
+      "loss": 0.2326,
+      "step": 3076
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028663760896637604,
+      "loss": 0.2727,
+      "step": 3077
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002866251556662515,
+      "loss": 0.5569,
+      "step": 3078
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000286612702366127,
+      "loss": 0.2849,
+      "step": 3079
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002866002490660025,
+      "loss": 0.4693,
+      "step": 3080
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028658779576587793,
+      "loss": 0.2493,
+      "step": 3081
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002865753424657534,
+      "loss": 0.2434,
+      "step": 3082
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028656288916562885,
+      "loss": 0.3003,
+      "step": 3083
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028655043586550433,
+      "loss": 0.4262,
+      "step": 3084
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002865379825653798,
+      "loss": 0.2847,
+      "step": 3085
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002865255292652553,
+      "loss": 0.4797,
+      "step": 3086
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028651307596513074,
+      "loss": 0.2227,
+      "step": 3087
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028650062266500617,
+      "loss": 0.3147,
+      "step": 3088
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028648816936488166,
+      "loss": 0.2713,
+      "step": 3089
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028647571606475714,
+      "loss": 0.3184,
+      "step": 3090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028646326276463263,
+      "loss": 0.384,
+      "step": 3091
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028645080946450806,
+      "loss": 0.296,
+      "step": 3092
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028643835616438355,
+      "loss": 0.2566,
+      "step": 3093
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000286425902864259,
+      "loss": 0.2936,
+      "step": 3094
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028641344956413446,
+      "loss": 0.2532,
+      "step": 3095
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028640099626400995,
+      "loss": 0.3298,
+      "step": 3096
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028638854296388544,
+      "loss": 1.1566,
+      "step": 3097
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028637608966376087,
+      "loss": 0.2463,
+      "step": 3098
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028636363636363636,
+      "loss": 0.3063,
+      "step": 3099
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002863511830635118,
+      "loss": 0.3289,
+      "step": 3100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002863387297633873,
+      "loss": 0.3083,
+      "step": 3101
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028632627646326276,
+      "loss": 0.4135,
+      "step": 3102
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028631382316313825,
+      "loss": 0.5522,
+      "step": 3103
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002863013698630137,
+      "loss": 0.2597,
+      "step": 3104
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002862889165628891,
+      "loss": 0.2642,
+      "step": 3105
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002862764632627646,
+      "loss": 0.2733,
+      "step": 3106
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002862640099626401,
+      "loss": 0.238,
+      "step": 3107
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028625155666251557,
+      "loss": 0.4381,
+      "step": 3108
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000286239103362391,
+      "loss": 0.5352,
+      "step": 3109
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002862266500622665,
+      "loss": 0.4666,
+      "step": 3110
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002862141967621419,
+      "loss": 0.3607,
+      "step": 3111
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002862017434620174,
+      "loss": 0.3244,
+      "step": 3112
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002861892901618929,
+      "loss": 0.3214,
+      "step": 3113
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002861768368617684,
+      "loss": 0.2894,
+      "step": 3114
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002861643835616438,
+      "loss": 0.358,
+      "step": 3115
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002861519302615193,
+      "loss": 0.4991,
+      "step": 3116
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028613947696139473,
+      "loss": 0.3377,
+      "step": 3117
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002861270236612702,
+      "loss": 0.2652,
+      "step": 3118
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002861145703611457,
+      "loss": 0.3389,
+      "step": 3119
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028610211706102113,
+      "loss": 0.2317,
+      "step": 3120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002860896637608966,
+      "loss": 0.3099,
+      "step": 3121
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028607721046077205,
+      "loss": 0.5493,
+      "step": 3122
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028606475716064754,
+      "loss": 0.2445,
+      "step": 3123
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000286052303860523,
+      "loss": 0.2574,
+      "step": 3124
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002860398505603985,
+      "loss": 0.2968,
+      "step": 3125
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028602739726027394,
+      "loss": 0.2285,
+      "step": 3126
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028601494396014943,
+      "loss": 0.3095,
+      "step": 3127
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028600249066002486,
+      "loss": 0.5345,
+      "step": 3128
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028599003735990035,
+      "loss": 0.3288,
+      "step": 3129
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028597758405977583,
+      "loss": 0.2551,
+      "step": 3130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002859651307596513,
+      "loss": 0.2268,
+      "step": 3131
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028595267745952675,
+      "loss": 0.3351,
+      "step": 3132
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028594022415940224,
+      "loss": 0.2718,
+      "step": 3133
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028592777085927767,
+      "loss": 0.5956,
+      "step": 3134
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028591531755915316,
+      "loss": 0.3158,
+      "step": 3135
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028590286425902864,
+      "loss": 0.2586,
+      "step": 3136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002858904109589041,
+      "loss": 0.2933,
+      "step": 3137
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028587795765877956,
+      "loss": 0.2481,
+      "step": 3138
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000285865504358655,
+      "loss": 0.2462,
+      "step": 3139
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002858530510585305,
+      "loss": 0.3062,
+      "step": 3140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028584059775840597,
+      "loss": 0.4389,
+      "step": 3141
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00028582814445828145,
+      "loss": 0.2967,
+      "step": 3142
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002858156911581569,
+      "loss": 0.2992,
+      "step": 3143
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028580323785803237,
+      "loss": 0.1817,
+      "step": 3144
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002857907845579078,
+      "loss": 0.2547,
+      "step": 3145
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002857783312577833,
+      "loss": 0.4176,
+      "step": 3146
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002857658779576588,
+      "loss": 0.5015,
+      "step": 3147
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028575342465753426,
+      "loss": 0.2195,
+      "step": 3148
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002857409713574097,
+      "loss": 0.2475,
+      "step": 3149
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002857285180572852,
+      "loss": 0.495,
+      "step": 3150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002857160647571606,
+      "loss": 0.2472,
+      "step": 3151
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002857036114570361,
+      "loss": 0.3346,
+      "step": 3152
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002856911581569116,
+      "loss": 0.4406,
+      "step": 3153
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000285678704856787,
+      "loss": 0.284,
+      "step": 3154
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002856662515566625,
+      "loss": 0.2343,
+      "step": 3155
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028565379825653793,
+      "loss": 0.2946,
+      "step": 3156
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002856413449564134,
+      "loss": 0.274,
+      "step": 3157
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002856288916562889,
+      "loss": 0.33,
+      "step": 3158
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002856164383561644,
+      "loss": 0.4515,
+      "step": 3159
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002856039850560398,
+      "loss": 0.2493,
+      "step": 3160
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002855915317559153,
+      "loss": 0.2348,
+      "step": 3161
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028557907845579074,
+      "loss": 0.2196,
+      "step": 3162
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028556662515566623,
+      "loss": 0.2661,
+      "step": 3163
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002855541718555417,
+      "loss": 0.2756,
+      "step": 3164
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002855417185554172,
+      "loss": 0.3668,
+      "step": 3165
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028552926525529264,
+      "loss": 0.3858,
+      "step": 3166
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028551681195516807,
+      "loss": 0.4441,
+      "step": 3167
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028550435865504355,
+      "loss": 0.2415,
+      "step": 3168
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028549190535491904,
+      "loss": 0.2227,
+      "step": 3169
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002854794520547945,
+      "loss": 0.2309,
+      "step": 3170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028546699875466996,
+      "loss": 0.2679,
+      "step": 3171
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028545454545454544,
+      "loss": 0.3353,
+      "step": 3172
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002854420921544209,
+      "loss": 0.2509,
+      "step": 3173
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028542963885429636,
+      "loss": 0.3903,
+      "step": 3174
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028541718555417185,
+      "loss": 0.3219,
+      "step": 3175
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002854047322540473,
+      "loss": 0.3074,
+      "step": 3176
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028539227895392277,
+      "loss": 0.3898,
+      "step": 3177
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028537982565379825,
+      "loss": 0.4674,
+      "step": 3178
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002853673723536737,
+      "loss": 0.2792,
+      "step": 3179
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028535491905354917,
+      "loss": 0.2545,
+      "step": 3180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028534246575342466,
+      "loss": 0.1933,
+      "step": 3181
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002853300124533001,
+      "loss": 0.284,
+      "step": 3182
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002853175591531756,
+      "loss": 0.3076,
+      "step": 3183
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000285305105853051,
+      "loss": 0.3042,
+      "step": 3184
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002852926525529265,
+      "loss": 0.2744,
+      "step": 3185
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000285280199252802,
+      "loss": 0.2865,
+      "step": 3186
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002852677459526774,
+      "loss": 0.2326,
+      "step": 3187
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002852552926525529,
+      "loss": 0.245,
+      "step": 3188
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002852428393524284,
+      "loss": 0.3167,
+      "step": 3189
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002852303860523038,
+      "loss": 0.2982,
+      "step": 3190
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002852179327521793,
+      "loss": 0.4861,
+      "step": 3191
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002852054794520548,
+      "loss": 0.2695,
+      "step": 3192
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002851930261519302,
+      "loss": 0.2849,
+      "step": 3193
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002851805728518057,
+      "loss": 0.2837,
+      "step": 3194
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002851681195516812,
+      "loss": 0.2333,
+      "step": 3195
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002851556662515566,
+      "loss": 0.4821,
+      "step": 3196
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002851432129514321,
+      "loss": 0.5539,
+      "step": 3197
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028513075965130755,
+      "loss": 0.2472,
+      "step": 3198
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028511830635118303,
+      "loss": 0.2437,
+      "step": 3199
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002851058530510585,
+      "loss": 0.2179,
+      "step": 3200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028509339975093395,
+      "loss": 0.3126,
+      "step": 3201
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028508094645080944,
+      "loss": 0.3625,
+      "step": 3202
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002850684931506849,
+      "loss": 0.5828,
+      "step": 3203
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028505603985056035,
+      "loss": 0.2415,
+      "step": 3204
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028504358655043584,
+      "loss": 0.3847,
+      "step": 3205
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028503113325031133,
+      "loss": 0.279,
+      "step": 3206
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028501867995018676,
+      "loss": 0.2569,
+      "step": 3207
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028500622665006225,
+      "loss": 0.2782,
+      "step": 3208
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002849937733499377,
+      "loss": 0.5007,
+      "step": 3209
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028498132004981316,
+      "loss": 0.3303,
+      "step": 3210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028496886674968865,
+      "loss": 0.2591,
+      "step": 3211
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028495641344956414,
+      "loss": 0.234,
+      "step": 3212
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028494396014943957,
+      "loss": 0.29,
+      "step": 3213
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028493150684931505,
+      "loss": 0.2912,
+      "step": 3214
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002849190535491905,
+      "loss": 0.473,
+      "step": 3215
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000284906600249066,
+      "loss": 0.626,
+      "step": 3216
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00028489414694894146,
+      "loss": 0.4791,
+      "step": 3217
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002848816936488169,
+      "loss": 0.3256,
+      "step": 3218
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002848692403486924,
+      "loss": 0.2531,
+      "step": 3219
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002848567870485678,
+      "loss": 0.3126,
+      "step": 3220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002848443337484433,
+      "loss": 0.4088,
+      "step": 3221
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002848318804483188,
+      "loss": 0.4289,
+      "step": 3222
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028481942714819427,
+      "loss": 0.3728,
+      "step": 3223
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002848069738480697,
+      "loss": 0.2784,
+      "step": 3224
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002847945205479452,
+      "loss": 0.2744,
+      "step": 3225
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002847820672478206,
+      "loss": 0.3204,
+      "step": 3226
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002847696139476961,
+      "loss": 0.2232,
+      "step": 3227
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002847571606475716,
+      "loss": 0.4293,
+      "step": 3228
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000284744707347447,
+      "loss": 0.3154,
+      "step": 3229
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002847322540473225,
+      "loss": 0.4383,
+      "step": 3230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028471980074719794,
+      "loss": 0.3128,
+      "step": 3231
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028470734744707343,
+      "loss": 0.2112,
+      "step": 3232
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002846948941469489,
+      "loss": 0.4104,
+      "step": 3233
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002846824408468244,
+      "loss": 0.5424,
+      "step": 3234
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028466998754669983,
+      "loss": 0.3294,
+      "step": 3235
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002846575342465753,
+      "loss": 0.287,
+      "step": 3236
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028464508094645075,
+      "loss": 0.2509,
+      "step": 3237
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028463262764632624,
+      "loss": 0.2859,
+      "step": 3238
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002846201743462017,
+      "loss": 0.2795,
+      "step": 3239
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002846077210460772,
+      "loss": 0.3022,
+      "step": 3240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028459526774595264,
+      "loss": 0.4314,
+      "step": 3241
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028458281444582813,
+      "loss": 0.2001,
+      "step": 3242
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028457036114570356,
+      "loss": 0.3202,
+      "step": 3243
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028455790784557905,
+      "loss": 0.2508,
+      "step": 3244
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028454545454545453,
+      "loss": 0.3504,
+      "step": 3245
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028453300124532996,
+      "loss": 0.3554,
+      "step": 3246
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00028452054794520545,
+      "loss": 0.5081,
+      "step": 3247
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002845080946450809,
+      "loss": 0.2812,
+      "step": 3248
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028449564134495637,
+      "loss": 0.3477,
+      "step": 3249
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028448318804483186,
+      "loss": 0.3895,
+      "step": 3250
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028447073474470734,
+      "loss": 0.2405,
+      "step": 3251
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002844582814445828,
+      "loss": 0.5195,
+      "step": 3252
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028444582814445826,
+      "loss": 0.3238,
+      "step": 3253
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002844333748443337,
+      "loss": 0.2884,
+      "step": 3254
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002844209215442092,
+      "loss": 0.3318,
+      "step": 3255
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028440846824408467,
+      "loss": 0.3306,
+      "step": 3256
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028439601494396015,
+      "loss": 0.248,
+      "step": 3257
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002843835616438356,
+      "loss": 0.3655,
+      "step": 3258
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028437110834371107,
+      "loss": 0.3905,
+      "step": 3259
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002843586550435865,
+      "loss": 0.2921,
+      "step": 3260
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000284346201743462,
+      "loss": 0.2426,
+      "step": 3261
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002843337484433375,
+      "loss": 0.2267,
+      "step": 3262
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002843212951432129,
+      "loss": 0.3707,
+      "step": 3263
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002843088418430884,
+      "loss": 0.2385,
+      "step": 3264
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002842963885429638,
+      "loss": 0.3317,
+      "step": 3265
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002842839352428393,
+      "loss": 0.3806,
+      "step": 3266
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002842714819427148,
+      "loss": 0.3388,
+      "step": 3267
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002842590286425903,
+      "loss": 0.2741,
+      "step": 3268
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002842465753424657,
+      "loss": 0.3013,
+      "step": 3269
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002842341220423412,
+      "loss": 0.3353,
+      "step": 3270
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028422166874221663,
+      "loss": 0.4299,
+      "step": 3271
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002842092154420921,
+      "loss": 0.7081,
+      "step": 3272
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002841967621419676,
+      "loss": 0.2773,
+      "step": 3273
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002841843088418431,
+      "loss": 0.2436,
+      "step": 3274
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002841718555417185,
+      "loss": 0.2768,
+      "step": 3275
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000284159402241594,
+      "loss": 0.2886,
+      "step": 3276
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028414694894146944,
+      "loss": 0.4185,
+      "step": 3277
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028413449564134493,
+      "loss": 0.5761,
+      "step": 3278
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002841220423412204,
+      "loss": 0.2832,
+      "step": 3279
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028410958904109585,
+      "loss": 0.2429,
+      "step": 3280
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028409713574097133,
+      "loss": 0.3589,
+      "step": 3281
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028408468244084677,
+      "loss": 0.3598,
+      "step": 3282
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028407222914072225,
+      "loss": 0.2674,
+      "step": 3283
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028405977584059774,
+      "loss": 0.5976,
+      "step": 3284
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002840473225404732,
+      "loss": 0.2713,
+      "step": 3285
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028403486924034866,
+      "loss": 0.296,
+      "step": 3286
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028402241594022414,
+      "loss": 0.4386,
+      "step": 3287
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002840099626400996,
+      "loss": 0.4705,
+      "step": 3288
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028399750933997506,
+      "loss": 0.2253,
+      "step": 3289
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028398505603985055,
+      "loss": 0.3588,
+      "step": 3290
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028397260273972603,
+      "loss": 0.4881,
+      "step": 3291
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028396014943960147,
+      "loss": 0.3254,
+      "step": 3292
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002839476961394769,
+      "loss": 0.2246,
+      "step": 3293
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002839352428393524,
+      "loss": 0.2519,
+      "step": 3294
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028392278953922787,
+      "loss": 0.2218,
+      "step": 3295
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028391033623910336,
+      "loss": 0.2043,
+      "step": 3296
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002838978829389788,
+      "loss": 0.4969,
+      "step": 3297
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002838854296388543,
+      "loss": 0.3577,
+      "step": 3298
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002838729763387297,
+      "loss": 0.2829,
+      "step": 3299
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002838605230386052,
+      "loss": 0.3107,
+      "step": 3300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002838480697384807,
+      "loss": 0.227,
+      "step": 3301
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028383561643835617,
+      "loss": 0.3224,
+      "step": 3302
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002838231631382316,
+      "loss": 0.4702,
+      "step": 3303
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002838107098381071,
+      "loss": 0.2798,
+      "step": 3304
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002837982565379825,
+      "loss": 0.2246,
+      "step": 3305
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000283785803237858,
+      "loss": 0.2097,
+      "step": 3306
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002837733499377335,
+      "loss": 0.2232,
+      "step": 3307
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002837608966376089,
+      "loss": 0.274,
+      "step": 3308
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002837484433374844,
+      "loss": 0.4915,
+      "step": 3309
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028373599003735984,
+      "loss": 0.2591,
+      "step": 3310
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002837235367372353,
+      "loss": 0.3467,
+      "step": 3311
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002837110834371108,
+      "loss": 0.2433,
+      "step": 3312
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002836986301369863,
+      "loss": 0.2949,
+      "step": 3313
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028368617683686173,
+      "loss": 0.4701,
+      "step": 3314
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002836737235367372,
+      "loss": 0.3095,
+      "step": 3315
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028366127023661265,
+      "loss": 0.5373,
+      "step": 3316
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028364881693648814,
+      "loss": 0.354,
+      "step": 3317
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002836363636363636,
+      "loss": 0.2509,
+      "step": 3318
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002836239103362391,
+      "loss": 0.2576,
+      "step": 3319
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028361145703611454,
+      "loss": 0.356,
+      "step": 3320
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028359900373599,
+      "loss": 0.292,
+      "step": 3321
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028358655043586546,
+      "loss": 0.3887,
+      "step": 3322
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028357409713574094,
+      "loss": 0.3921,
+      "step": 3323
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028356164383561643,
+      "loss": 0.317,
+      "step": 3324
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028354919053549186,
+      "loss": 0.3201,
+      "step": 3325
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00028353673723536735,
+      "loss": 0.2688,
+      "step": 3326
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002835242839352428,
+      "loss": 0.2179,
+      "step": 3327
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028351183063511827,
+      "loss": 0.4262,
+      "step": 3328
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028349937733499375,
+      "loss": 0.2745,
+      "step": 3329
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028348692403486924,
+      "loss": 0.2899,
+      "step": 3330
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028347447073474467,
+      "loss": 0.1989,
+      "step": 3331
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028346201743462016,
+      "loss": 0.5255,
+      "step": 3332
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002834495641344956,
+      "loss": 0.221,
+      "step": 3333
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002834371108343711,
+      "loss": 0.5203,
+      "step": 3334
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028342465753424656,
+      "loss": 0.2534,
+      "step": 3335
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028341220423412205,
+      "loss": 0.3466,
+      "step": 3336
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002833997509339975,
+      "loss": 0.2769,
+      "step": 3337
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028338729763387297,
+      "loss": 0.2467,
+      "step": 3338
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002833748443337484,
+      "loss": 0.4266,
+      "step": 3339
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002833623910336239,
+      "loss": 0.4267,
+      "step": 3340
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028334993773349937,
+      "loss": 0.4044,
+      "step": 3341
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002833374844333748,
+      "loss": 0.9393,
+      "step": 3342
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002833250311332503,
+      "loss": 0.3061,
+      "step": 3343
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002833125778331257,
+      "loss": 0.2408,
+      "step": 3344
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002833001245330012,
+      "loss": 0.3478,
+      "step": 3345
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002832876712328767,
+      "loss": 0.3502,
+      "step": 3346
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002832752179327522,
+      "loss": 0.6704,
+      "step": 3347
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002832627646326276,
+      "loss": 0.2607,
+      "step": 3348
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002832503113325031,
+      "loss": 0.2113,
+      "step": 3349
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028323785803237853,
+      "loss": 0.193,
+      "step": 3350
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000283225404732254,
+      "loss": 0.2291,
+      "step": 3351
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002832129514321295,
+      "loss": 0.3733,
+      "step": 3352
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000283200498132005,
+      "loss": 0.4662,
+      "step": 3353
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002831880448318804,
+      "loss": 0.3424,
+      "step": 3354
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028317559153175585,
+      "loss": 0.2359,
+      "step": 3355
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028316313823163134,
+      "loss": 0.1904,
+      "step": 3356
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028315068493150683,
+      "loss": 0.3482,
+      "step": 3357
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002831382316313823,
+      "loss": 0.3402,
+      "step": 3358
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028312577833125775,
+      "loss": 0.4132,
+      "step": 3359
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028311332503113323,
+      "loss": 0.2656,
+      "step": 3360
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028310087173100866,
+      "loss": 0.8662,
+      "step": 3361
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028308841843088415,
+      "loss": 0.2499,
+      "step": 3362
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028307596513075964,
+      "loss": 0.2309,
+      "step": 3363
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002830635118306351,
+      "loss": 0.3542,
+      "step": 3364
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028305105853051055,
+      "loss": 0.3368,
+      "step": 3365
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028303860523038604,
+      "loss": 0.456,
+      "step": 3366
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002830261519302615,
+      "loss": 0.3464,
+      "step": 3367
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028301369863013696,
+      "loss": 0.3503,
+      "step": 3368
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028300124533001245,
+      "loss": 0.2365,
+      "step": 3369
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028298879202988793,
+      "loss": 0.3761,
+      "step": 3370
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028297633872976336,
+      "loss": 0.2877,
+      "step": 3371
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002829638854296388,
+      "loss": 0.6987,
+      "step": 3372
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002829514321295143,
+      "loss": 0.2673,
+      "step": 3373
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028293897882938977,
+      "loss": 0.3041,
+      "step": 3374
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00028292652552926525,
+      "loss": 0.2054,
+      "step": 3375
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002829140722291407,
+      "loss": 0.3548,
+      "step": 3376
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002829016189290162,
+      "loss": 0.3438,
+      "step": 3377
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002828891656288916,
+      "loss": 0.6918,
+      "step": 3378
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002828767123287671,
+      "loss": 0.2977,
+      "step": 3379
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002828642590286426,
+      "loss": 0.2867,
+      "step": 3380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028285180572851806,
+      "loss": 0.2526,
+      "step": 3381
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002828393524283935,
+      "loss": 0.2144,
+      "step": 3382
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000282826899128269,
+      "loss": 0.2747,
+      "step": 3383
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002828144458281444,
+      "loss": 0.5469,
+      "step": 3384
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002828019925280199,
+      "loss": 0.2759,
+      "step": 3385
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002827895392278954,
+      "loss": 0.3276,
+      "step": 3386
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002827770859277708,
+      "loss": 0.3152,
+      "step": 3387
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002827646326276463,
+      "loss": 0.4245,
+      "step": 3388
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028275217932752174,
+      "loss": 0.2331,
+      "step": 3389
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002827397260273972,
+      "loss": 0.2681,
+      "step": 3390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002827272727272727,
+      "loss": 0.5094,
+      "step": 3391
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002827148194271482,
+      "loss": 0.2534,
+      "step": 3392
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028270236612702363,
+      "loss": 0.2345,
+      "step": 3393
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002826899128268991,
+      "loss": 0.4424,
+      "step": 3394
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028267745952677455,
+      "loss": 0.2391,
+      "step": 3395
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028266500622665003,
+      "loss": 0.2906,
+      "step": 3396
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002826525529265255,
+      "loss": 0.6366,
+      "step": 3397
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000282640099626401,
+      "loss": 0.3817,
+      "step": 3398
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028262764632627644,
+      "loss": 0.2989,
+      "step": 3399
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002826151930261519,
+      "loss": 0.3446,
+      "step": 3400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028260273972602736,
+      "loss": 0.3886,
+      "step": 3401
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028259028642590284,
+      "loss": 0.495,
+      "step": 3402
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028257783312577833,
+      "loss": 0.6428,
+      "step": 3403
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00028256537982565376,
+      "loss": 0.3523,
+      "step": 3404
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028255292652552925,
+      "loss": 0.3174,
+      "step": 3405
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002825404732254047,
+      "loss": 0.2935,
+      "step": 3406
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028252801992528017,
+      "loss": 0.2038,
+      "step": 3407
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028251556662515565,
+      "loss": 0.2631,
+      "step": 3408
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028250311332503114,
+      "loss": 0.4087,
+      "step": 3409
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028249066002490657,
+      "loss": 0.2633,
+      "step": 3410
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028247820672478206,
+      "loss": 0.2577,
+      "step": 3411
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002824657534246575,
+      "loss": 0.2943,
+      "step": 3412
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000282453300124533,
+      "loss": 0.3161,
+      "step": 3413
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028244084682440846,
+      "loss": 0.2742,
+      "step": 3414
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028242839352428395,
+      "loss": 0.3167,
+      "step": 3415
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002824159402241594,
+      "loss": 0.5281,
+      "step": 3416
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028240348692403487,
+      "loss": 0.2627,
+      "step": 3417
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002823910336239103,
+      "loss": 0.2555,
+      "step": 3418
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002823785803237858,
+      "loss": 0.232,
+      "step": 3419
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028236612702366127,
+      "loss": 0.3133,
+      "step": 3420
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002823536737235367,
+      "loss": 0.3754,
+      "step": 3421
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002823412204234122,
+      "loss": 0.611,
+      "step": 3422
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002823287671232876,
+      "loss": 0.2343,
+      "step": 3423
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002823163138231631,
+      "loss": 0.5003,
+      "step": 3424
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002823038605230386,
+      "loss": 0.2611,
+      "step": 3425
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002822914072229141,
+      "loss": 0.2368,
+      "step": 3426
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002822789539227895,
+      "loss": 0.4825,
+      "step": 3427
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000282266500622665,
+      "loss": 0.4183,
+      "step": 3428
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00028225404732254043,
+      "loss": 0.2602,
+      "step": 3429
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002822415940224159,
+      "loss": 0.2984,
+      "step": 3430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002822291407222914,
+      "loss": 0.2445,
+      "step": 3431
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002822166874221669,
+      "loss": 0.2675,
+      "step": 3432
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002822042341220423,
+      "loss": 0.3012,
+      "step": 3433
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028219178082191775,
+      "loss": 0.4456,
+      "step": 3434
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028217932752179324,
+      "loss": 0.2879,
+      "step": 3435
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002821668742216687,
+      "loss": 0.2443,
+      "step": 3436
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002821544209215442,
+      "loss": 0.2286,
+      "step": 3437
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028214196762141964,
+      "loss": 0.2634,
+      "step": 3438
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028212951432129513,
+      "loss": 0.2551,
+      "step": 3439
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028211706102117056,
+      "loss": 0.3189,
+      "step": 3440
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028210460772104605,
+      "loss": 0.3701,
+      "step": 3441
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028209215442092153,
+      "loss": 0.4032,
+      "step": 3442
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000282079701120797,
+      "loss": 0.2598,
+      "step": 3443
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028206724782067245,
+      "loss": 0.3038,
+      "step": 3444
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028205479452054794,
+      "loss": 0.2824,
+      "step": 3445
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028204234122042337,
+      "loss": 0.4351,
+      "step": 3446
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028202988792029886,
+      "loss": 0.5681,
+      "step": 3447
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028201743462017434,
+      "loss": 0.34,
+      "step": 3448
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002820049813200498,
+      "loss": 0.2807,
+      "step": 3449
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028199252801992526,
+      "loss": 0.189,
+      "step": 3450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002819800747198007,
+      "loss": 0.2127,
+      "step": 3451
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002819676214196762,
+      "loss": 0.2541,
+      "step": 3452
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028195516811955167,
+      "loss": 0.4593,
+      "step": 3453
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028194271481942715,
+      "loss": 0.3083,
+      "step": 3454
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002819302615193026,
+      "loss": 0.2233,
+      "step": 3455
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028191780821917807,
+      "loss": 0.352,
+      "step": 3456
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002819053549190535,
+      "loss": 0.3133,
+      "step": 3457
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000281892901618929,
+      "loss": 0.2503,
+      "step": 3458
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002818804483188045,
+      "loss": 0.8345,
+      "step": 3459
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028186799501867996,
+      "loss": 0.2687,
+      "step": 3460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002818555417185554,
+      "loss": 0.3195,
+      "step": 3461
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002818430884184309,
+      "loss": 0.2249,
+      "step": 3462
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002818306351183063,
+      "loss": 0.2269,
+      "step": 3463
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002818181818181818,
+      "loss": 0.4683,
+      "step": 3464
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002818057285180573,
+      "loss": 0.3395,
+      "step": 3465
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002817932752179327,
+      "loss": 0.4297,
+      "step": 3466
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002817808219178082,
+      "loss": 0.3741,
+      "step": 3467
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028176836861768364,
+      "loss": 0.2561,
+      "step": 3468
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002817559153175591,
+      "loss": 0.3697,
+      "step": 3469
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002817434620174346,
+      "loss": 0.3308,
+      "step": 3470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002817310087173101,
+      "loss": 0.315,
+      "step": 3471
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002817185554171855,
+      "loss": 0.4201,
+      "step": 3472
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000281706102117061,
+      "loss": 0.275,
+      "step": 3473
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028169364881693644,
+      "loss": 0.2409,
+      "step": 3474
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028168119551681193,
+      "loss": 0.2828,
+      "step": 3475
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002816687422166874,
+      "loss": 0.308,
+      "step": 3476
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002816562889165629,
+      "loss": 0.2495,
+      "step": 3477
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028164383561643834,
+      "loss": 0.4373,
+      "step": 3478
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002816313823163138,
+      "loss": 0.2525,
+      "step": 3479
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028161892901618925,
+      "loss": 0.2965,
+      "step": 3480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028160647571606474,
+      "loss": 0.2808,
+      "step": 3481
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002815940224159402,
+      "loss": 0.2733,
+      "step": 3482
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028158156911581566,
+      "loss": 0.3908,
+      "step": 3483
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028156911581569114,
+      "loss": 0.5071,
+      "step": 3484
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002815566625155666,
+      "loss": 0.3446,
+      "step": 3485
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028154420921544206,
+      "loss": 0.2083,
+      "step": 3486
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028153175591531755,
+      "loss": 0.2583,
+      "step": 3487
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028151930261519304,
+      "loss": 0.2436,
+      "step": 3488
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028150684931506847,
+      "loss": 0.3877,
+      "step": 3489
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028149439601494395,
+      "loss": 0.2797,
+      "step": 3490
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002814819427148194,
+      "loss": 0.4336,
+      "step": 3491
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028146948941469487,
+      "loss": 0.3267,
+      "step": 3492
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028145703611457036,
+      "loss": 0.2219,
+      "step": 3493
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028144458281444584,
+      "loss": 0.2477,
+      "step": 3494
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002814321295143213,
+      "loss": 0.2602,
+      "step": 3495
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002814196762141967,
+      "loss": 0.3274,
+      "step": 3496
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002814072229140722,
+      "loss": 0.3744,
+      "step": 3497
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002813947696139477,
+      "loss": 0.5302,
+      "step": 3498
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028138231631382317,
+      "loss": 0.2599,
+      "step": 3499
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002813698630136986,
+      "loss": 0.2787,
+      "step": 3500
+    },
+    {
+      "epoch": 1.34,
+      "eval_cer": 0.1155003219072517,
+      "eval_loss": 0.3954097628593445,
+      "eval_runtime": 46.661,
+      "eval_samples_per_second": 37.569,
+      "eval_steps_per_second": 4.715,
+      "eval_wer": 0.20898764538172604,
+      "step": 3500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002813574097135741,
+      "loss": 0.2147,
+      "step": 3501
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002813449564134495,
+      "loss": 0.2195,
+      "step": 3502
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000281332503113325,
+      "loss": 0.6195,
+      "step": 3503
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002813200498132005,
+      "loss": 0.3297,
+      "step": 3504
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000281307596513076,
+      "loss": 0.3818,
+      "step": 3505
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002812951432129514,
+      "loss": 0.2387,
+      "step": 3506
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002812826899128269,
+      "loss": 0.3068,
+      "step": 3507
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00028127023661270233,
+      "loss": 0.3216,
+      "step": 3508
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002812577833125778,
+      "loss": 0.3536,
+      "step": 3509
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002812453300124533,
+      "loss": 0.2913,
+      "step": 3510
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028123287671232873,
+      "loss": 0.3375,
+      "step": 3511
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002812204234122042,
+      "loss": 0.3066,
+      "step": 3512
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028120797011207965,
+      "loss": 0.277,
+      "step": 3513
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028119551681195514,
+      "loss": 0.304,
+      "step": 3514
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002811830635118306,
+      "loss": 0.4731,
+      "step": 3515
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002811706102117061,
+      "loss": 0.4346,
+      "step": 3516
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028115815691158154,
+      "loss": 0.2653,
+      "step": 3517
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028114570361145703,
+      "loss": 0.243,
+      "step": 3518
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028113325031133246,
+      "loss": 0.2722,
+      "step": 3519
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028112079701120795,
+      "loss": 0.2652,
+      "step": 3520
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028110834371108343,
+      "loss": 0.2895,
+      "step": 3521
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028109589041095886,
+      "loss": 0.4241,
+      "step": 3522
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028108343711083435,
+      "loss": 0.2444,
+      "step": 3523
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028107098381070984,
+      "loss": 0.2463,
+      "step": 3524
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028105853051058527,
+      "loss": 0.2757,
+      "step": 3525
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028104607721046075,
+      "loss": 0.2406,
+      "step": 3526
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028103362391033624,
+      "loss": 0.3071,
+      "step": 3527
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002810211706102117,
+      "loss": 0.4139,
+      "step": 3528
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028100871731008716,
+      "loss": 0.2723,
+      "step": 3529
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002809962640099626,
+      "loss": 0.1928,
+      "step": 3530
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002809838107098381,
+      "loss": 0.2543,
+      "step": 3531
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028097135740971356,
+      "loss": 0.2264,
+      "step": 3532
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000280958904109589,
+      "loss": 0.5093,
+      "step": 3533
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002809464508094645,
+      "loss": 0.547,
+      "step": 3534
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028093399750933997,
+      "loss": 0.218,
+      "step": 3535
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002809215442092154,
+      "loss": 0.3129,
+      "step": 3536
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002809090909090909,
+      "loss": 0.2421,
+      "step": 3537
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002808966376089664,
+      "loss": 0.2902,
+      "step": 3538
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002808841843088418,
+      "loss": 0.339,
+      "step": 3539
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002808717310087173,
+      "loss": 0.4214,
+      "step": 3540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002808592777085928,
+      "loss": 0.4009,
+      "step": 3541
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002808468244084682,
+      "loss": 0.4605,
+      "step": 3542
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002808343711083437,
+      "loss": 0.3566,
+      "step": 3543
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028082191780821913,
+      "loss": 0.3904,
+      "step": 3544
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002808094645080946,
+      "loss": 0.2985,
+      "step": 3545
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002807970112079701,
+      "loss": 0.2822,
+      "step": 3546
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028078455790784553,
+      "loss": 0.4112,
+      "step": 3547
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.000280772104607721,
+      "loss": 0.796,
+      "step": 3548
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028075965130759645,
+      "loss": 0.2768,
+      "step": 3549
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028074719800747194,
+      "loss": 0.2924,
+      "step": 3550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002807347447073474,
+      "loss": 0.4938,
+      "step": 3551
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002807222914072229,
+      "loss": 0.2844,
+      "step": 3552
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028070983810709834,
+      "loss": 0.6272,
+      "step": 3553
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028069738480697383,
+      "loss": 0.3154,
+      "step": 3554
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028068493150684926,
+      "loss": 0.4723,
+      "step": 3555
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028067247820672475,
+      "loss": 0.3029,
+      "step": 3556
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028066002490660023,
+      "loss": 0.2327,
+      "step": 3557
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002806475716064757,
+      "loss": 0.3865,
+      "step": 3558
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028063511830635115,
+      "loss": 0.5413,
+      "step": 3559
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002806226650062266,
+      "loss": 0.2584,
+      "step": 3560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028061021170610207,
+      "loss": 0.2683,
+      "step": 3561
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00028059775840597756,
+      "loss": 0.2571,
+      "step": 3562
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00028058530510585304,
+      "loss": 0.4376,
+      "step": 3563
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002805728518057285,
+      "loss": 0.3684,
+      "step": 3564
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00028056039850560396,
+      "loss": 0.4355,
+      "step": 3565
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002805479452054794,
+      "loss": 0.5171,
+      "step": 3566
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002805354919053549,
+      "loss": 0.377,
+      "step": 3567
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00028052303860523037,
+      "loss": 0.3176,
+      "step": 3568
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00028051058530510585,
+      "loss": 0.2654,
+      "step": 3569
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002804981320049813,
+      "loss": 0.3876,
+      "step": 3570
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00028048567870485677,
+      "loss": 0.266,
+      "step": 3571
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002804732254047322,
+      "loss": 0.4587,
+      "step": 3572
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002804607721046077,
+      "loss": 0.2362,
+      "step": 3573
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002804483188044832,
+      "loss": 0.2314,
+      "step": 3574
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002804358655043586,
+      "loss": 0.307,
+      "step": 3575
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002804234122042341,
+      "loss": 0.2051,
+      "step": 3576
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002804109589041095,
+      "loss": 0.4043,
+      "step": 3577
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000280398505603985,
+      "loss": 0.5107,
+      "step": 3578
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002803860523038605,
+      "loss": 0.3416,
+      "step": 3579
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000280373599003736,
+      "loss": 0.2954,
+      "step": 3580
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002803611457036114,
+      "loss": 0.2199,
+      "step": 3581
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002803486924034869,
+      "loss": 0.3246,
+      "step": 3582
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00028033623910336233,
+      "loss": 0.2517,
+      "step": 3583
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002803237858032378,
+      "loss": 0.5089,
+      "step": 3584
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002803113325031133,
+      "loss": 0.2693,
+      "step": 3585
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002802988792029888,
+      "loss": 0.2467,
+      "step": 3586
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002802864259028642,
+      "loss": 0.2477,
+      "step": 3587
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002802739726027397,
+      "loss": 0.2291,
+      "step": 3588
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028026151930261514,
+      "loss": 0.3128,
+      "step": 3589
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028024906600249063,
+      "loss": 0.4842,
+      "step": 3590
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002802366127023661,
+      "loss": 0.4248,
+      "step": 3591
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028022415940224155,
+      "loss": 0.4202,
+      "step": 3592
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028021170610211703,
+      "loss": 0.4074,
+      "step": 3593
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028019925280199247,
+      "loss": 0.2597,
+      "step": 3594
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028018679950186795,
+      "loss": 0.3986,
+      "step": 3595
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028017434620174344,
+      "loss": 0.3717,
+      "step": 3596
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002801618929016189,
+      "loss": 0.5486,
+      "step": 3597
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028014943960149436,
+      "loss": 0.2786,
+      "step": 3598
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028013698630136984,
+      "loss": 0.3227,
+      "step": 3599
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002801245330012453,
+      "loss": 0.2698,
+      "step": 3600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028011207970112076,
+      "loss": 0.2962,
+      "step": 3601
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028009962640099625,
+      "loss": 0.3684,
+      "step": 3602
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028008717310087173,
+      "loss": 0.5815,
+      "step": 3603
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028007471980074717,
+      "loss": 0.3371,
+      "step": 3604
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028006226650062265,
+      "loss": 0.3302,
+      "step": 3605
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002800498132004981,
+      "loss": 0.2768,
+      "step": 3606
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028003735990037357,
+      "loss": 0.2048,
+      "step": 3607
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028002490660024906,
+      "loss": 0.3159,
+      "step": 3608
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002800124533001245,
+      "loss": 0.5228,
+      "step": 3609
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028,
+      "loss": 0.3469,
+      "step": 3610
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002799875466998754,
+      "loss": 0.2103,
+      "step": 3611
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002799750933997509,
+      "loss": 0.264,
+      "step": 3612
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002799626400996264,
+      "loss": 0.2559,
+      "step": 3613
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00027995018679950187,
+      "loss": 0.2769,
+      "step": 3614
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002799377334993773,
+      "loss": 0.3605,
+      "step": 3615
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002799252801992528,
+      "loss": 0.4254,
+      "step": 3616
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002799128268991282,
+      "loss": 0.3033,
+      "step": 3617
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002799003735990037,
+      "loss": 0.2373,
+      "step": 3618
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002798879202988792,
+      "loss": 0.2732,
+      "step": 3619
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002798754669987547,
+      "loss": 0.3979,
+      "step": 3620
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002798630136986301,
+      "loss": 0.3541,
+      "step": 3621
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00027985056039850554,
+      "loss": 0.4606,
+      "step": 3622
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000279838107098381,
+      "loss": 0.2926,
+      "step": 3623
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002798256537982565,
+      "loss": 0.2536,
+      "step": 3624
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000279813200498132,
+      "loss": 0.2206,
+      "step": 3625
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00027980074719800743,
+      "loss": 0.202,
+      "step": 3626
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002797882938978829,
+      "loss": 0.2194,
+      "step": 3627
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00027977584059775835,
+      "loss": 0.3931,
+      "step": 3628
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00027976338729763384,
+      "loss": 0.2771,
+      "step": 3629
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002797509339975093,
+      "loss": 0.2727,
+      "step": 3630
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002797384806973848,
+      "loss": 0.2461,
+      "step": 3631
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00027972602739726024,
+      "loss": 0.2463,
+      "step": 3632
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002797135740971357,
+      "loss": 0.2685,
+      "step": 3633
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00027970112079701116,
+      "loss": 0.4782,
+      "step": 3634
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00027968866749688664,
+      "loss": 0.457,
+      "step": 3635
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00027967621419676213,
+      "loss": 0.2994,
+      "step": 3636
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002796637608966376,
+      "loss": 0.2369,
+      "step": 3637
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00027965130759651305,
+      "loss": 0.3428,
+      "step": 3638
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002796388542963885,
+      "loss": 0.3598,
+      "step": 3639
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00027962640099626397,
+      "loss": 0.3508,
+      "step": 3640
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00027961394769613945,
+      "loss": 0.4008,
+      "step": 3641
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00027960149439601494,
+      "loss": 0.3274,
+      "step": 3642
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00027958904109589037,
+      "loss": 0.2722,
+      "step": 3643
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00027957658779576586,
+      "loss": 0.4094,
+      "step": 3644
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002795641344956413,
+      "loss": 0.2478,
+      "step": 3645
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002795516811955168,
+      "loss": 0.3638,
+      "step": 3646
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00027953922789539226,
+      "loss": 0.4487,
+      "step": 3647
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00027952677459526775,
+      "loss": 0.2509,
+      "step": 3648
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002795143212951432,
+      "loss": 0.2804,
+      "step": 3649
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00027950186799501867,
+      "loss": 0.3744,
+      "step": 3650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002794894146948941,
+      "loss": 0.232,
+      "step": 3651
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002794769613947696,
+      "loss": 0.339,
+      "step": 3652
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00027946450809464507,
+      "loss": 0.5169,
+      "step": 3653
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002794520547945205,
+      "loss": 0.3303,
+      "step": 3654
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000279439601494396,
+      "loss": 0.2364,
+      "step": 3655
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002794271481942714,
+      "loss": 0.2644,
+      "step": 3656
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002794146948941469,
+      "loss": 0.3518,
+      "step": 3657
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002794022415940224,
+      "loss": 0.2356,
+      "step": 3658
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002793897882938979,
+      "loss": 0.4591,
+      "step": 3659
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002793773349937733,
+      "loss": 0.2569,
+      "step": 3660
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002793648816936488,
+      "loss": 0.402,
+      "step": 3661
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00027935242839352423,
+      "loss": 0.2988,
+      "step": 3662
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002793399750933997,
+      "loss": 0.2594,
+      "step": 3663
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002793275217932752,
+      "loss": 0.246,
+      "step": 3664
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002793150684931507,
+      "loss": 0.3561,
+      "step": 3665
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002793026151930261,
+      "loss": 0.3588,
+      "step": 3666
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002792901618929016,
+      "loss": 0.2532,
+      "step": 3667
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027927770859277704,
+      "loss": 0.3826,
+      "step": 3668
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027926525529265253,
+      "loss": 0.2357,
+      "step": 3669
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000279252801992528,
+      "loss": 0.3079,
+      "step": 3670
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027924034869240345,
+      "loss": 0.6591,
+      "step": 3671
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027922789539227893,
+      "loss": 0.5611,
+      "step": 3672
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027921544209215436,
+      "loss": 0.2541,
+      "step": 3673
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027920298879202985,
+      "loss": 0.3409,
+      "step": 3674
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027919053549190534,
+      "loss": 0.2608,
+      "step": 3675
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002791780821917808,
+      "loss": 0.2126,
+      "step": 3676
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027916562889165625,
+      "loss": 0.3045,
+      "step": 3677
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027915317559153174,
+      "loss": 0.3529,
+      "step": 3678
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002791407222914072,
+      "loss": 0.4011,
+      "step": 3679
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027912826899128266,
+      "loss": 0.2538,
+      "step": 3680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027911581569115815,
+      "loss": 0.3123,
+      "step": 3681
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027910336239103363,
+      "loss": 0.3443,
+      "step": 3682
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027909090909090906,
+      "loss": 0.2508,
+      "step": 3683
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027907845579078455,
+      "loss": 0.5128,
+      "step": 3684
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027906600249066,
+      "loss": 0.3757,
+      "step": 3685
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027905354919053547,
+      "loss": 0.2586,
+      "step": 3686
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00027904109589041095,
+      "loss": 0.3549,
+      "step": 3687
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002790286425902864,
+      "loss": 0.2918,
+      "step": 3688
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002790161892901619,
+      "loss": 0.3527,
+      "step": 3689
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002790037359900373,
+      "loss": 0.5148,
+      "step": 3690
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002789912826899128,
+      "loss": 0.5118,
+      "step": 3691
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002789788293897883,
+      "loss": 0.2792,
+      "step": 3692
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00027896637608966376,
+      "loss": 0.2562,
+      "step": 3693
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002789539227895392,
+      "loss": 0.2693,
+      "step": 3694
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002789414694894147,
+      "loss": 0.3127,
+      "step": 3695
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002789290161892901,
+      "loss": 0.3149,
+      "step": 3696
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002789165628891656,
+      "loss": 0.3985,
+      "step": 3697
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002789041095890411,
+      "loss": 0.2025,
+      "step": 3698
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002788916562889166,
+      "loss": 0.9434,
+      "step": 3699
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000278879202988792,
+      "loss": 0.2569,
+      "step": 3700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00027886674968866744,
+      "loss": 0.3004,
+      "step": 3701
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002788542963885429,
+      "loss": 0.2883,
+      "step": 3702
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002788418430884184,
+      "loss": 0.4935,
+      "step": 3703
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002788293897882939,
+      "loss": 0.2786,
+      "step": 3704
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00027881693648816933,
+      "loss": 0.3949,
+      "step": 3705
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002788044831880448,
+      "loss": 0.3354,
+      "step": 3706
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00027879202988792025,
+      "loss": 0.2398,
+      "step": 3707
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00027877957658779573,
+      "loss": 0.3795,
+      "step": 3708
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002787671232876712,
+      "loss": 0.7635,
+      "step": 3709
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002787546699875467,
+      "loss": 0.3069,
+      "step": 3710
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00027874221668742214,
+      "loss": 0.2547,
+      "step": 3711
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002787297633872976,
+      "loss": 0.307,
+      "step": 3712
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00027871731008717306,
+      "loss": 0.3431,
+      "step": 3713
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00027870485678704854,
+      "loss": 0.3424,
+      "step": 3714
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00027869240348692403,
+      "loss": 0.3763,
+      "step": 3715
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00027867995018679946,
+      "loss": 0.5909,
+      "step": 3716
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00027866749688667495,
+      "loss": 0.2526,
+      "step": 3717
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002786550435865504,
+      "loss": 0.5692,
+      "step": 3718
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00027864259028642587,
+      "loss": 0.292,
+      "step": 3719
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00027863013698630135,
+      "loss": 0.2969,
+      "step": 3720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00027861768368617684,
+      "loss": 0.3639,
+      "step": 3721
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00027860523038605227,
+      "loss": 0.5696,
+      "step": 3722
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00027859277708592776,
+      "loss": 0.3599,
+      "step": 3723
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002785803237858032,
+      "loss": 0.3999,
+      "step": 3724
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002785678704856787,
+      "loss": 0.326,
+      "step": 3725
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00027855541718555416,
+      "loss": 0.2733,
+      "step": 3726
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00027854296388542965,
+      "loss": 0.3095,
+      "step": 3727
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002785305105853051,
+      "loss": 0.4634,
+      "step": 3728
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00027851805728518057,
+      "loss": 0.2513,
+      "step": 3729
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000278505603985056,
+      "loss": 0.2926,
+      "step": 3730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002784931506849315,
+      "loss": 0.2612,
+      "step": 3731
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00027848069738480697,
+      "loss": 0.3545,
+      "step": 3732
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002784682440846824,
+      "loss": 0.5004,
+      "step": 3733
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002784557907845579,
+      "loss": 0.6157,
+      "step": 3734
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002784433374844333,
+      "loss": 0.6597,
+      "step": 3735
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002784308841843088,
+      "loss": 0.3088,
+      "step": 3736
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002784184308841843,
+      "loss": 0.2766,
+      "step": 3737
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002784059775840598,
+      "loss": 0.2752,
+      "step": 3738
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002783935242839352,
+      "loss": 0.3213,
+      "step": 3739
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002783810709838107,
+      "loss": 0.2879,
+      "step": 3740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00027836861768368613,
+      "loss": 0.6876,
+      "step": 3741
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002783561643835616,
+      "loss": 0.2805,
+      "step": 3742
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002783437110834371,
+      "loss": 0.3263,
+      "step": 3743
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002783312577833126,
+      "loss": 0.2249,
+      "step": 3744
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000278318804483188,
+      "loss": 0.3469,
+      "step": 3745
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002783063511830635,
+      "loss": 0.3153,
+      "step": 3746
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027829389788293894,
+      "loss": 0.3707,
+      "step": 3747
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002782814445828144,
+      "loss": 0.3331,
+      "step": 3748
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002782689912826899,
+      "loss": 0.2324,
+      "step": 3749
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027825653798256534,
+      "loss": 0.3428,
+      "step": 3750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027824408468244083,
+      "loss": 0.2893,
+      "step": 3751
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027823163138231626,
+      "loss": 0.3002,
+      "step": 3752
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027821917808219175,
+      "loss": 0.4227,
+      "step": 3753
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027820672478206723,
+      "loss": 0.2644,
+      "step": 3754
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002781942714819427,
+      "loss": 0.2229,
+      "step": 3755
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027818181818181815,
+      "loss": 0.3423,
+      "step": 3756
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027816936488169364,
+      "loss": 0.272,
+      "step": 3757
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027815691158156907,
+      "loss": 0.2164,
+      "step": 3758
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027814445828144456,
+      "loss": 0.4831,
+      "step": 3759
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027813200498132004,
+      "loss": 0.3038,
+      "step": 3760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027811955168119553,
+      "loss": 0.2953,
+      "step": 3761
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027810709838107096,
+      "loss": 0.3548,
+      "step": 3762
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002780946450809464,
+      "loss": 0.3772,
+      "step": 3763
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002780821917808219,
+      "loss": 0.2187,
+      "step": 3764
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027806973848069737,
+      "loss": 0.3856,
+      "step": 3765
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027805728518057285,
+      "loss": 0.4637,
+      "step": 3766
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002780448318804483,
+      "loss": 0.2461,
+      "step": 3767
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00027803237858032377,
+      "loss": 0.2914,
+      "step": 3768
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002780199252801992,
+      "loss": 0.3427,
+      "step": 3769
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002780074719800747,
+      "loss": 0.4177,
+      "step": 3770
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002779950186799502,
+      "loss": 0.289,
+      "step": 3771
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00027798256537982566,
+      "loss": 0.4921,
+      "step": 3772
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002779701120797011,
+      "loss": 0.2646,
+      "step": 3773
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002779576587795766,
+      "loss": 0.2358,
+      "step": 3774
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000277945205479452,
+      "loss": 0.247,
+      "step": 3775
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002779327521793275,
+      "loss": 0.2726,
+      "step": 3776
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000277920298879203,
+      "loss": 0.3647,
+      "step": 3777
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00027790784557907847,
+      "loss": 0.6624,
+      "step": 3778
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002778953922789539,
+      "loss": 0.2658,
+      "step": 3779
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00027788293897882934,
+      "loss": 0.241,
+      "step": 3780
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002778704856787048,
+      "loss": 0.2623,
+      "step": 3781
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002778580323785803,
+      "loss": 0.2706,
+      "step": 3782
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002778455790784558,
+      "loss": 0.3038,
+      "step": 3783
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002778331257783312,
+      "loss": 0.6302,
+      "step": 3784
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002778206724782067,
+      "loss": 0.2943,
+      "step": 3785
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00027780821917808214,
+      "loss": 0.4342,
+      "step": 3786
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00027779576587795763,
+      "loss": 0.2099,
+      "step": 3787
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002777833125778331,
+      "loss": 0.2,
+      "step": 3788
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002777708592777086,
+      "loss": 0.2679,
+      "step": 3789
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00027775840597758404,
+      "loss": 0.3823,
+      "step": 3790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002777459526774595,
+      "loss": 0.3946,
+      "step": 3791
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00027773349937733495,
+      "loss": 0.2722,
+      "step": 3792
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00027772104607721044,
+      "loss": 0.3147,
+      "step": 3793
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002777085927770859,
+      "loss": 0.2415,
+      "step": 3794
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00027769613947696136,
+      "loss": 0.3567,
+      "step": 3795
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00027768368617683684,
+      "loss": 0.4011,
+      "step": 3796
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002776712328767123,
+      "loss": 0.3912,
+      "step": 3797
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00027765877957658776,
+      "loss": 0.2695,
+      "step": 3798
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00027764632627646325,
+      "loss": 0.2373,
+      "step": 3799
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00027763387297633874,
+      "loss": 0.2247,
+      "step": 3800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00027762141967621417,
+      "loss": 0.2724,
+      "step": 3801
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00027760896637608965,
+      "loss": 0.3495,
+      "step": 3802
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002775965130759651,
+      "loss": 0.5747,
+      "step": 3803
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00027758405977584057,
+      "loss": 0.2606,
+      "step": 3804
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00027757160647571606,
+      "loss": 0.4155,
+      "step": 3805
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00027755915317559154,
+      "loss": 0.4604,
+      "step": 3806
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000277546699875467,
+      "loss": 0.2776,
+      "step": 3807
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00027753424657534246,
+      "loss": 0.248,
+      "step": 3808
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002775217932752179,
+      "loss": 0.4096,
+      "step": 3809
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002775093399750934,
+      "loss": 0.3171,
+      "step": 3810
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00027749688667496887,
+      "loss": 0.3556,
+      "step": 3811
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002774844333748443,
+      "loss": 0.2953,
+      "step": 3812
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002774719800747198,
+      "loss": 0.2315,
+      "step": 3813
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002774595267745952,
+      "loss": 0.2848,
+      "step": 3814
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002774470734744707,
+      "loss": 0.3677,
+      "step": 3815
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002774346201743462,
+      "loss": 0.4549,
+      "step": 3816
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002774221668742217,
+      "loss": 0.3826,
+      "step": 3817
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002774097135740971,
+      "loss": 0.2894,
+      "step": 3818
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002773972602739726,
+      "loss": 0.3217,
+      "step": 3819
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00027738480697384803,
+      "loss": 0.2668,
+      "step": 3820
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002773723536737235,
+      "loss": 0.2985,
+      "step": 3821
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000277359900373599,
+      "loss": 0.4664,
+      "step": 3822
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002773474470734745,
+      "loss": 0.3329,
+      "step": 3823
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002773349937733499,
+      "loss": 0.3027,
+      "step": 3824
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002773225404732254,
+      "loss": 0.3325,
+      "step": 3825
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027731008717310084,
+      "loss": 0.2159,
+      "step": 3826
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002772976338729763,
+      "loss": 0.2037,
+      "step": 3827
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002772851805728518,
+      "loss": 0.5508,
+      "step": 3828
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027727272727272724,
+      "loss": 0.3774,
+      "step": 3829
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027726027397260273,
+      "loss": 0.3421,
+      "step": 3830
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027724782067247816,
+      "loss": 0.3368,
+      "step": 3831
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027723536737235365,
+      "loss": 0.4191,
+      "step": 3832
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027722291407222913,
+      "loss": 0.4418,
+      "step": 3833
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002772104607721046,
+      "loss": 0.4389,
+      "step": 3834
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027719800747198005,
+      "loss": 0.3308,
+      "step": 3835
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027718555417185554,
+      "loss": 0.2485,
+      "step": 3836
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027717310087173097,
+      "loss": 0.2478,
+      "step": 3837
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027716064757160645,
+      "loss": 0.2731,
+      "step": 3838
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027714819427148194,
+      "loss": 0.3498,
+      "step": 3839
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027713574097135743,
+      "loss": 0.3723,
+      "step": 3840
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027712328767123286,
+      "loss": 0.396,
+      "step": 3841
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002771108343711083,
+      "loss": 0.3059,
+      "step": 3842
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002770983810709838,
+      "loss": 0.4458,
+      "step": 3843
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027708592777085926,
+      "loss": 0.2692,
+      "step": 3844
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027707347447073475,
+      "loss": 0.3736,
+      "step": 3845
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002770610211706102,
+      "loss": 0.2582,
+      "step": 3846
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00027704856787048567,
+      "loss": 0.5981,
+      "step": 3847
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002770361145703611,
+      "loss": 0.2019,
+      "step": 3848
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002770236612702366,
+      "loss": 0.3299,
+      "step": 3849
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002770112079701121,
+      "loss": 0.2631,
+      "step": 3850
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00027699875466998756,
+      "loss": 0.2363,
+      "step": 3851
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000276986301369863,
+      "loss": 0.4031,
+      "step": 3852
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002769738480697385,
+      "loss": 0.5108,
+      "step": 3853
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002769613947696139,
+      "loss": 0.3068,
+      "step": 3854
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002769489414694894,
+      "loss": 0.3138,
+      "step": 3855
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002769364881693649,
+      "loss": 0.246,
+      "step": 3856
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002769240348692403,
+      "loss": 0.3212,
+      "step": 3857
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002769115815691158,
+      "loss": 0.2139,
+      "step": 3858
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00027689912826899123,
+      "loss": 0.4629,
+      "step": 3859
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002768866749688667,
+      "loss": 0.5555,
+      "step": 3860
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002768742216687422,
+      "loss": 0.32,
+      "step": 3861
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00027686176836861764,
+      "loss": 0.2561,
+      "step": 3862
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002768493150684931,
+      "loss": 0.2169,
+      "step": 3863
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002768368617683686,
+      "loss": 0.2408,
+      "step": 3864
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00027682440846824404,
+      "loss": 0.3971,
+      "step": 3865
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00027682440846824404,
+      "loss": 1.8002,
+      "step": 3866
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00027681195516811953,
+      "loss": 0.3098,
+      "step": 3867
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000276799501867995,
+      "loss": 0.26,
+      "step": 3868
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00027678704856787045,
+      "loss": 0.3309,
+      "step": 3869
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00027677459526774593,
+      "loss": 0.2795,
+      "step": 3870
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002767621419676214,
+      "loss": 0.2712,
+      "step": 3871
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00027674968866749685,
+      "loss": 0.2952,
+      "step": 3872
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00027673723536737234,
+      "loss": 0.2571,
+      "step": 3873
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00027672478206724777,
+      "loss": 0.232,
+      "step": 3874
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00027671232876712326,
+      "loss": 0.2295,
+      "step": 3875
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00027669987546699874,
+      "loss": 0.2059,
+      "step": 3876
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002766874221668742,
+      "loss": 0.2713,
+      "step": 3877
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00027667496886674966,
+      "loss": 0.5524,
+      "step": 3878
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00027666251556662515,
+      "loss": 0.2508,
+      "step": 3879
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002766500622665006,
+      "loss": 0.2694,
+      "step": 3880
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00027663760896637607,
+      "loss": 0.3048,
+      "step": 3881
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00027662515566625155,
+      "loss": 0.3202,
+      "step": 3882
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000276612702366127,
+      "loss": 0.2506,
+      "step": 3883
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00027660024906600247,
+      "loss": 0.5679,
+      "step": 3884
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002765877957658779,
+      "loss": 0.567,
+      "step": 3885
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002765753424657534,
+      "loss": 0.282,
+      "step": 3886
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002765628891656289,
+      "loss": 0.2569,
+      "step": 3887
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00027655043586550436,
+      "loss": 0.3166,
+      "step": 3888
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002765379825653798,
+      "loss": 0.2352,
+      "step": 3889
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002765255292652553,
+      "loss": 0.2957,
+      "step": 3890
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002765130759651307,
+      "loss": 0.4057,
+      "step": 3891
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002765006226650062,
+      "loss": 0.2252,
+      "step": 3892
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002764881693648817,
+      "loss": 0.2371,
+      "step": 3893
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002764757160647571,
+      "loss": 0.3453,
+      "step": 3894
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002764632627646326,
+      "loss": 0.2835,
+      "step": 3895
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00027645080946450803,
+      "loss": 0.4585,
+      "step": 3896
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002764383561643835,
+      "loss": 0.56,
+      "step": 3897
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000276425902864259,
+      "loss": 0.2748,
+      "step": 3898
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002764134495641345,
+      "loss": 0.4714,
+      "step": 3899
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002764009962640099,
+      "loss": 0.2335,
+      "step": 3900
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002763885429638854,
+      "loss": 0.2433,
+      "step": 3901
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027637608966376084,
+      "loss": 0.305,
+      "step": 3902
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027636363636363633,
+      "loss": 0.3896,
+      "step": 3903
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002763511830635118,
+      "loss": 0.2626,
+      "step": 3904
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002763387297633873,
+      "loss": 0.2016,
+      "step": 3905
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027632627646326273,
+      "loss": 0.2581,
+      "step": 3906
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027631382316313817,
+      "loss": 0.2375,
+      "step": 3907
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027630136986301365,
+      "loss": 0.2905,
+      "step": 3908
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027628891656288914,
+      "loss": 0.5388,
+      "step": 3909
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002762764632627646,
+      "loss": 0.2924,
+      "step": 3910
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027626400996264006,
+      "loss": 0.2036,
+      "step": 3911
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027625155666251554,
+      "loss": 0.2595,
+      "step": 3912
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000276239103362391,
+      "loss": 0.2731,
+      "step": 3913
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027622665006226646,
+      "loss": 0.3332,
+      "step": 3914
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027621419676214195,
+      "loss": 0.3471,
+      "step": 3915
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027620174346201743,
+      "loss": 0.3664,
+      "step": 3916
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027618929016189287,
+      "loss": 0.2741,
+      "step": 3917
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027617683686176835,
+      "loss": 0.2332,
+      "step": 3918
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002761643835616438,
+      "loss": 0.2421,
+      "step": 3919
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027615193026151927,
+      "loss": 0.2632,
+      "step": 3920
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00027613947696139476,
+      "loss": 0.3325,
+      "step": 3921
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002761270236612702,
+      "loss": 0.3553,
+      "step": 3922
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002761145703611457,
+      "loss": 0.2457,
+      "step": 3923
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002761021170610211,
+      "loss": 0.2336,
+      "step": 3924
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002760896637608966,
+      "loss": 0.1964,
+      "step": 3925
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002760772104607721,
+      "loss": 0.4867,
+      "step": 3926
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00027606475716064757,
+      "loss": 0.2633,
+      "step": 3927
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000276052303860523,
+      "loss": 0.5642,
+      "step": 3928
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002760398505603985,
+      "loss": 0.2416,
+      "step": 3929
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002760273972602739,
+      "loss": 0.2166,
+      "step": 3930
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002760149439601494,
+      "loss": 0.2414,
+      "step": 3931
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002760024906600249,
+      "loss": 0.3552,
+      "step": 3932
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002759900373599004,
+      "loss": 0.3136,
+      "step": 3933
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002759775840597758,
+      "loss": 0.4726,
+      "step": 3934
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002759651307596513,
+      "loss": 0.2962,
+      "step": 3935
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002759526774595267,
+      "loss": 0.2416,
+      "step": 3936
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002759402241594022,
+      "loss": 0.2063,
+      "step": 3937
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002759277708592777,
+      "loss": 0.2884,
+      "step": 3938
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00027591531755915313,
+      "loss": 0.5304,
+      "step": 3939
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002759028642590286,
+      "loss": 0.2542,
+      "step": 3940
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00027589041095890405,
+      "loss": 0.3726,
+      "step": 3941
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00027587795765877954,
+      "loss": 0.2803,
+      "step": 3942
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000275865504358655,
+      "loss": 0.2824,
+      "step": 3943
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002758530510585305,
+      "loss": 0.3288,
+      "step": 3944
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00027584059775840594,
+      "loss": 0.2582,
+      "step": 3945
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002758281444582814,
+      "loss": 0.2764,
+      "step": 3946
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00027581569115815686,
+      "loss": 0.3318,
+      "step": 3947
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00027580323785803234,
+      "loss": 0.265,
+      "step": 3948
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00027579078455790783,
+      "loss": 0.281,
+      "step": 3949
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002757783312577833,
+      "loss": 0.2177,
+      "step": 3950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00027576587795765875,
+      "loss": 0.3064,
+      "step": 3951
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00027575342465753424,
+      "loss": 0.2746,
+      "step": 3952
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00027574097135740967,
+      "loss": 0.6876,
+      "step": 3953
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00027572851805728515,
+      "loss": 0.3125,
+      "step": 3954
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00027571606475716064,
+      "loss": 0.2362,
+      "step": 3955
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00027570361145703607,
+      "loss": 0.188,
+      "step": 3956
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00027569115815691156,
+      "loss": 0.2302,
+      "step": 3957
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000275678704856787,
+      "loss": 0.2581,
+      "step": 3958
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002756662515566625,
+      "loss": 0.5068,
+      "step": 3959
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00027565379825653796,
+      "loss": 0.299,
+      "step": 3960
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00027564134495641345,
+      "loss": 0.2643,
+      "step": 3961
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002756288916562889,
+      "loss": 0.1827,
+      "step": 3962
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00027561643835616437,
+      "loss": 0.3701,
+      "step": 3963
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002756039850560398,
+      "loss": 0.2187,
+      "step": 3964
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002755915317559153,
+      "loss": 0.3525,
+      "step": 3965
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00027557907845579077,
+      "loss": 0.4284,
+      "step": 3966
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00027556662515566626,
+      "loss": 0.2296,
+      "step": 3967
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002755541718555417,
+      "loss": 0.2409,
+      "step": 3968
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002755417185554171,
+      "loss": 0.3706,
+      "step": 3969
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002755292652552926,
+      "loss": 0.2599,
+      "step": 3970
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002755168119551681,
+      "loss": 0.3123,
+      "step": 3971
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002755043586550436,
+      "loss": 0.4648,
+      "step": 3972
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000275491905354919,
+      "loss": 0.2861,
+      "step": 3973
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002754794520547945,
+      "loss": 0.2139,
+      "step": 3974
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00027546699875466993,
+      "loss": 0.2368,
+      "step": 3975
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002754545454545454,
+      "loss": 0.2516,
+      "step": 3976
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002754420921544209,
+      "loss": 0.3313,
+      "step": 3977
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002754296388542964,
+      "loss": 0.4857,
+      "step": 3978
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002754171855541718,
+      "loss": 0.2947,
+      "step": 3979
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002754047322540473,
+      "loss": 0.2447,
+      "step": 3980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027539227895392274,
+      "loss": 0.2446,
+      "step": 3981
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027537982565379823,
+      "loss": 0.2802,
+      "step": 3982
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002753673723536737,
+      "loss": 0.1835,
+      "step": 3983
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027535491905354915,
+      "loss": 0.7385,
+      "step": 3984
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027534246575342463,
+      "loss": 0.2184,
+      "step": 3985
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027533001245330006,
+      "loss": 0.5491,
+      "step": 3986
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027531755915317555,
+      "loss": 0.2714,
+      "step": 3987
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027530510585305104,
+      "loss": 0.2533,
+      "step": 3988
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002752926525529265,
+      "loss": 0.2456,
+      "step": 3989
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027528019925280195,
+      "loss": 0.3319,
+      "step": 3990
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027526774595267744,
+      "loss": 0.4997,
+      "step": 3991
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002752552926525529,
+      "loss": 0.3448,
+      "step": 3992
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027524283935242836,
+      "loss": 0.3038,
+      "step": 3993
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027523038605230385,
+      "loss": 0.2635,
+      "step": 3994
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027521793275217933,
+      "loss": 0.2435,
+      "step": 3995
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027520547945205476,
+      "loss": 0.2959,
+      "step": 3996
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027519302615193025,
+      "loss": 0.329,
+      "step": 3997
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002751805728518057,
+      "loss": 0.3117,
+      "step": 3998
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027516811955168117,
+      "loss": 0.3211,
+      "step": 3999
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00027515566625155665,
+      "loss": 0.2775,
+      "step": 4000
+    },
+    {
+      "epoch": 1.53,
+      "eval_cer": 0.11333476403222975,
+      "eval_loss": 0.39375272393226624,
+      "eval_runtime": 47.0547,
+      "eval_samples_per_second": 37.255,
+      "eval_steps_per_second": 4.675,
+      "eval_wer": 0.19921256279132915,
+      "step": 4000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002751432129514321,
+      "loss": 0.2663,
+      "step": 4001
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002751307596513076,
+      "loss": 0.3908,
+      "step": 4002
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000275118306351183,
+      "loss": 0.5007,
+      "step": 4003
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002751058530510585,
+      "loss": 0.2583,
+      "step": 4004
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000275093399750934,
+      "loss": 0.3619,
+      "step": 4005
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00027508094645080946,
+      "loss": 0.2404,
+      "step": 4006
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002750684931506849,
+      "loss": 0.2524,
+      "step": 4007
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002750560398505604,
+      "loss": 0.3323,
+      "step": 4008
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002750435865504358,
+      "loss": 0.5721,
+      "step": 4009
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002750311332503113,
+      "loss": 0.2279,
+      "step": 4010
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002750186799501868,
+      "loss": 0.318,
+      "step": 4011
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002750062266500623,
+      "loss": 0.3365,
+      "step": 4012
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002749937733499377,
+      "loss": 0.2703,
+      "step": 4013
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002749813200498132,
+      "loss": 0.2349,
+      "step": 4014
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002749688667496886,
+      "loss": 0.4048,
+      "step": 4015
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002749564134495641,
+      "loss": 0.3652,
+      "step": 4016
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002749439601494396,
+      "loss": 0.7341,
+      "step": 4017
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00027493150684931503,
+      "loss": 0.235,
+      "step": 4018
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002749190535491905,
+      "loss": 0.3427,
+      "step": 4019
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00027490660024906595,
+      "loss": 0.281,
+      "step": 4020
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00027489414694894143,
+      "loss": 0.3417,
+      "step": 4021
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002748816936488169,
+      "loss": 0.4911,
+      "step": 4022
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002748692403486924,
+      "loss": 0.254,
+      "step": 4023
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00027485678704856784,
+      "loss": 0.2112,
+      "step": 4024
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002748443337484433,
+      "loss": 0.2449,
+      "step": 4025
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00027483188044831876,
+      "loss": 0.2516,
+      "step": 4026
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00027481942714819424,
+      "loss": 0.2271,
+      "step": 4027
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00027480697384806973,
+      "loss": 0.5627,
+      "step": 4028
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002747945205479452,
+      "loss": 0.283,
+      "step": 4029
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00027478206724782065,
+      "loss": 0.3804,
+      "step": 4030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002747696139476961,
+      "loss": 0.2685,
+      "step": 4031
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00027475716064757157,
+      "loss": 0.3793,
+      "step": 4032
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00027474470734744705,
+      "loss": 0.3834,
+      "step": 4033
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00027473225404732254,
+      "loss": 0.3683,
+      "step": 4034
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00027471980074719797,
+      "loss": 0.2552,
+      "step": 4035
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00027470734744707346,
+      "loss": 0.1796,
+      "step": 4036
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002746948941469489,
+      "loss": 0.4086,
+      "step": 4037
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002746824408468244,
+      "loss": 0.5459,
+      "step": 4038
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00027466998754669986,
+      "loss": 0.2553,
+      "step": 4039
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00027465753424657535,
+      "loss": 0.3662,
+      "step": 4040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002746450809464508,
+      "loss": 0.3741,
+      "step": 4041
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00027463262764632627,
+      "loss": 0.3034,
+      "step": 4042
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002746201743462017,
+      "loss": 0.381,
+      "step": 4043
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002746077210460772,
+      "loss": 0.5985,
+      "step": 4044
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00027459526774595267,
+      "loss": 0.1938,
+      "step": 4045
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00027458281444582816,
+      "loss": 0.3288,
+      "step": 4046
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002745703611457036,
+      "loss": 0.4542,
+      "step": 4047
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000274557907845579,
+      "loss": 0.4702,
+      "step": 4048
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002745454545454545,
+      "loss": 0.2307,
+      "step": 4049
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00027453300124533,
+      "loss": 0.2181,
+      "step": 4050
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002745205479452055,
+      "loss": 0.2433,
+      "step": 4051
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002745080946450809,
+      "loss": 0.2491,
+      "step": 4052
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002744956413449564,
+      "loss": 0.3705,
+      "step": 4053
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00027448318804483183,
+      "loss": 0.3203,
+      "step": 4054
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002744707347447073,
+      "loss": 0.1828,
+      "step": 4055
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002744582814445828,
+      "loss": 0.3322,
+      "step": 4056
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002744458281444583,
+      "loss": 0.345,
+      "step": 4057
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002744333748443337,
+      "loss": 0.3795,
+      "step": 4058
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002744209215442092,
+      "loss": 0.4694,
+      "step": 4059
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027440846824408464,
+      "loss": 0.2783,
+      "step": 4060
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002743960149439601,
+      "loss": 0.2908,
+      "step": 4061
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002743835616438356,
+      "loss": 0.2091,
+      "step": 4062
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027437110834371104,
+      "loss": 0.1943,
+      "step": 4063
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027435865504358653,
+      "loss": 0.3658,
+      "step": 4064
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027434620174346196,
+      "loss": 0.2713,
+      "step": 4065
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027433374844333745,
+      "loss": 0.4756,
+      "step": 4066
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027432129514321293,
+      "loss": 0.2752,
+      "step": 4067
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002743088418430884,
+      "loss": 0.3599,
+      "step": 4068
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027429638854296385,
+      "loss": 0.232,
+      "step": 4069
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027428393524283934,
+      "loss": 0.5552,
+      "step": 4070
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027427148194271477,
+      "loss": 0.2921,
+      "step": 4071
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027425902864259026,
+      "loss": 0.7816,
+      "step": 4072
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027424657534246574,
+      "loss": 0.5543,
+      "step": 4073
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027423412204234123,
+      "loss": 0.3323,
+      "step": 4074
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027422166874221666,
+      "loss": 0.2757,
+      "step": 4075
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027420921544209215,
+      "loss": 0.4615,
+      "step": 4076
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002741967621419676,
+      "loss": 0.4492,
+      "step": 4077
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027418430884184307,
+      "loss": 0.7914,
+      "step": 4078
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027417185554171855,
+      "loss": 0.2494,
+      "step": 4079
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000274159402241594,
+      "loss": 0.3073,
+      "step": 4080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00027414694894146947,
+      "loss": 0.2988,
+      "step": 4081
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002741344956413449,
+      "loss": 0.2421,
+      "step": 4082
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002741220423412204,
+      "loss": 0.3064,
+      "step": 4083
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002741095890410959,
+      "loss": 0.6716,
+      "step": 4084
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00027409713574097136,
+      "loss": 0.269,
+      "step": 4085
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002740846824408468,
+      "loss": 0.2314,
+      "step": 4086
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002740722291407223,
+      "loss": 0.237,
+      "step": 4087
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002740597758405977,
+      "loss": 0.3148,
+      "step": 4088
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002740473225404732,
+      "loss": 0.2036,
+      "step": 4089
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002740348692403487,
+      "loss": 0.3451,
+      "step": 4090
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00027402241594022417,
+      "loss": 0.3868,
+      "step": 4091
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002740099626400996,
+      "loss": 0.3779,
+      "step": 4092
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002739975093399751,
+      "loss": 0.3107,
+      "step": 4093
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002739850560398505,
+      "loss": 0.5151,
+      "step": 4094
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000273972602739726,
+      "loss": 0.413,
+      "step": 4095
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002739601494396015,
+      "loss": 0.3889,
+      "step": 4096
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002739476961394769,
+      "loss": 0.5136,
+      "step": 4097
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002739352428393524,
+      "loss": 0.359,
+      "step": 4098
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00027392278953922784,
+      "loss": 0.2301,
+      "step": 4099
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00027391033623910333,
+      "loss": 0.3183,
+      "step": 4100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002738978829389788,
+      "loss": 0.31,
+      "step": 4101
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002738854296388543,
+      "loss": 0.2931,
+      "step": 4102
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00027387297633872974,
+      "loss": 0.6498,
+      "step": 4103
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002738605230386052,
+      "loss": 0.2882,
+      "step": 4104
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00027384806973848065,
+      "loss": 0.2237,
+      "step": 4105
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00027383561643835614,
+      "loss": 0.3224,
+      "step": 4106
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002738231631382316,
+      "loss": 0.2718,
+      "step": 4107
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002738107098381071,
+      "loss": 0.2778,
+      "step": 4108
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00027379825653798254,
+      "loss": 0.5685,
+      "step": 4109
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000273785803237858,
+      "loss": 0.2817,
+      "step": 4110
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00027377334993773346,
+      "loss": 0.2416,
+      "step": 4111
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00027376089663760895,
+      "loss": 0.2705,
+      "step": 4112
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00027374844333748444,
+      "loss": 0.2284,
+      "step": 4113
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00027373599003735987,
+      "loss": 0.3033,
+      "step": 4114
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00027372353673723535,
+      "loss": 0.3559,
+      "step": 4115
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002737110834371108,
+      "loss": 0.4815,
+      "step": 4116
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00027369863013698627,
+      "loss": 0.3237,
+      "step": 4117
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00027368617683686176,
+      "loss": 0.2144,
+      "step": 4118
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00027367372353673724,
+      "loss": 0.3721,
+      "step": 4119
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002736612702366127,
+      "loss": 0.39,
+      "step": 4120
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00027364881693648816,
+      "loss": 0.3424,
+      "step": 4121
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002736363636363636,
+      "loss": 0.4502,
+      "step": 4122
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002736239103362391,
+      "loss": 0.2328,
+      "step": 4123
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00027361145703611457,
+      "loss": 0.3335,
+      "step": 4124
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00027359900373599005,
+      "loss": 0.2497,
+      "step": 4125
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002735865504358655,
+      "loss": 0.2746,
+      "step": 4126
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002735740971357409,
+      "loss": 0.3741,
+      "step": 4127
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002735616438356164,
+      "loss": 0.4703,
+      "step": 4128
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002735491905354919,
+      "loss": 0.3736,
+      "step": 4129
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002735367372353674,
+      "loss": 0.378,
+      "step": 4130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002735242839352428,
+      "loss": 0.2675,
+      "step": 4131
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002735118306351183,
+      "loss": 0.2787,
+      "step": 4132
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00027349937733499373,
+      "loss": 0.2366,
+      "step": 4133
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002734869240348692,
+      "loss": 0.6257,
+      "step": 4134
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002734744707347447,
+      "loss": 0.4063,
+      "step": 4135
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002734620174346202,
+      "loss": 0.2569,
+      "step": 4136
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002734495641344956,
+      "loss": 0.3393,
+      "step": 4137
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002734371108343711,
+      "loss": 0.2733,
+      "step": 4138
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027342465753424654,
+      "loss": 0.251,
+      "step": 4139
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000273412204234122,
+      "loss": 0.3514,
+      "step": 4140
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002733997509339975,
+      "loss": 0.3593,
+      "step": 4141
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027338729763387294,
+      "loss": 0.2466,
+      "step": 4142
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027337484433374843,
+      "loss": 0.27,
+      "step": 4143
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027336239103362386,
+      "loss": 0.2805,
+      "step": 4144
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027334993773349935,
+      "loss": 0.3816,
+      "step": 4145
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027333748443337483,
+      "loss": 0.2218,
+      "step": 4146
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002733250311332503,
+      "loss": 0.502,
+      "step": 4147
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027331257783312575,
+      "loss": 0.2634,
+      "step": 4148
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027330012453300124,
+      "loss": 0.2208,
+      "step": 4149
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027328767123287667,
+      "loss": 0.2969,
+      "step": 4150
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027327521793275215,
+      "loss": 0.2887,
+      "step": 4151
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027326276463262764,
+      "loss": 0.2579,
+      "step": 4152
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027325031133250313,
+      "loss": 0.3837,
+      "step": 4153
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027323785803237856,
+      "loss": 0.3017,
+      "step": 4154
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027322540473225405,
+      "loss": 0.3488,
+      "step": 4155
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002732129514321295,
+      "loss": 0.3049,
+      "step": 4156
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027320049813200496,
+      "loss": 0.1905,
+      "step": 4157
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027318804483188045,
+      "loss": 0.3797,
+      "step": 4158
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002731755915317559,
+      "loss": 0.3393,
+      "step": 4159
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00027316313823163137,
+      "loss": 0.2671,
+      "step": 4160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002731506849315068,
+      "loss": 0.2937,
+      "step": 4161
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002731382316313823,
+      "loss": 0.3978,
+      "step": 4162
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002731257783312578,
+      "loss": 0.3163,
+      "step": 4163
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00027311332503113326,
+      "loss": 0.2709,
+      "step": 4164
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002731008717310087,
+      "loss": 0.2685,
+      "step": 4165
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002730884184308842,
+      "loss": 0.3046,
+      "step": 4166
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002730759651307596,
+      "loss": 0.6957,
+      "step": 4167
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002730635118306351,
+      "loss": 0.2368,
+      "step": 4168
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002730510585305106,
+      "loss": 0.3306,
+      "step": 4169
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00027303860523038607,
+      "loss": 0.309,
+      "step": 4170
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002730261519302615,
+      "loss": 0.3347,
+      "step": 4171
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000273013698630137,
+      "loss": 0.394,
+      "step": 4172
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002730012453300124,
+      "loss": 0.2822,
+      "step": 4173
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002729887920298879,
+      "loss": 0.3352,
+      "step": 4174
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002729763387297634,
+      "loss": 0.2157,
+      "step": 4175
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002729638854296388,
+      "loss": 0.3031,
+      "step": 4176
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002729514321295143,
+      "loss": 0.2912,
+      "step": 4177
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00027293897882938974,
+      "loss": 0.4943,
+      "step": 4178
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00027292652552926523,
+      "loss": 0.271,
+      "step": 4179
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002729140722291407,
+      "loss": 0.2212,
+      "step": 4180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002729016189290162,
+      "loss": 0.1964,
+      "step": 4181
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00027288916562889163,
+      "loss": 0.2784,
+      "step": 4182
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002728767123287671,
+      "loss": 0.2412,
+      "step": 4183
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00027286425902864255,
+      "loss": 0.357,
+      "step": 4184
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00027285180572851804,
+      "loss": 0.5017,
+      "step": 4185
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002728393524283935,
+      "loss": 0.2661,
+      "step": 4186
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00027282689912826896,
+      "loss": 0.2277,
+      "step": 4187
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00027281444582814444,
+      "loss": 0.3191,
+      "step": 4188
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002728019925280199,
+      "loss": 0.4673,
+      "step": 4189
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00027278953922789536,
+      "loss": 0.4562,
+      "step": 4190
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00027277708592777085,
+      "loss": 0.5206,
+      "step": 4191
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00027276463262764633,
+      "loss": 0.2892,
+      "step": 4192
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00027275217932752177,
+      "loss": 0.2798,
+      "step": 4193
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00027273972602739725,
+      "loss": 0.2282,
+      "step": 4194
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002727272727272727,
+      "loss": 0.2936,
+      "step": 4195
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00027271481942714817,
+      "loss": 0.3638,
+      "step": 4196
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00027270236612702366,
+      "loss": 0.5986,
+      "step": 4197
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002726899128268991,
+      "loss": 0.242,
+      "step": 4198
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002726774595267746,
+      "loss": 0.3061,
+      "step": 4199
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00027266500622665006,
+      "loss": 0.3542,
+      "step": 4200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002726525529265255,
+      "loss": 0.2237,
+      "step": 4201
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000272640099626401,
+      "loss": 0.3014,
+      "step": 4202
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00027262764632627647,
+      "loss": 0.5163,
+      "step": 4203
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002726151930261519,
+      "loss": 0.3648,
+      "step": 4204
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002726027397260274,
+      "loss": 0.2598,
+      "step": 4205
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002725902864259028,
+      "loss": 0.3121,
+      "step": 4206
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002725778331257783,
+      "loss": 0.2235,
+      "step": 4207
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002725653798256538,
+      "loss": 0.5142,
+      "step": 4208
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002725529265255292,
+      "loss": 0.4128,
+      "step": 4209
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002725404732254047,
+      "loss": 0.2648,
+      "step": 4210
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002725280199252802,
+      "loss": 0.3096,
+      "step": 4211
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002725155666251556,
+      "loss": 0.3401,
+      "step": 4212
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002725031133250311,
+      "loss": 0.2533,
+      "step": 4213
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002724906600249066,
+      "loss": 0.2424,
+      "step": 4214
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027247820672478203,
+      "loss": 0.4337,
+      "step": 4215
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002724657534246575,
+      "loss": 0.579,
+      "step": 4216
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000272453300124533,
+      "loss": 0.245,
+      "step": 4217
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027244084682440843,
+      "loss": 0.2885,
+      "step": 4218
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002724283935242839,
+      "loss": 0.2806,
+      "step": 4219
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027241594022415935,
+      "loss": 0.2729,
+      "step": 4220
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027240348692403484,
+      "loss": 0.2925,
+      "step": 4221
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002723910336239103,
+      "loss": 0.4414,
+      "step": 4222
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027237858032378576,
+      "loss": 0.335,
+      "step": 4223
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027236612702366124,
+      "loss": 0.2214,
+      "step": 4224
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027235367372353673,
+      "loss": 0.2336,
+      "step": 4225
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027234122042341216,
+      "loss": 0.2928,
+      "step": 4226
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027232876712328765,
+      "loss": 0.2125,
+      "step": 4227
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027231631382316313,
+      "loss": 0.365,
+      "step": 4228
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027230386052303857,
+      "loss": 0.6017,
+      "step": 4229
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027229140722291405,
+      "loss": 0.2439,
+      "step": 4230
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002722789539227895,
+      "loss": 0.2658,
+      "step": 4231
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027226650062266497,
+      "loss": 0.2683,
+      "step": 4232
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027225404732254046,
+      "loss": 0.2312,
+      "step": 4233
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027224159402241594,
+      "loss": 0.714,
+      "step": 4234
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002722291407222914,
+      "loss": 0.2733,
+      "step": 4235
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002722166874221668,
+      "loss": 0.2963,
+      "step": 4236
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002722042341220423,
+      "loss": 0.2708,
+      "step": 4237
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002721917808219178,
+      "loss": 0.2585,
+      "step": 4238
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00027217932752179327,
+      "loss": 0.2324,
+      "step": 4239
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002721668742216687,
+      "loss": 0.2259,
+      "step": 4240
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002721544209215442,
+      "loss": 0.4126,
+      "step": 4241
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002721419676214196,
+      "loss": 0.3112,
+      "step": 4242
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002721295143212951,
+      "loss": 0.2488,
+      "step": 4243
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002721170610211706,
+      "loss": 0.3077,
+      "step": 4244
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002721046077210461,
+      "loss": 0.312,
+      "step": 4245
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002720921544209215,
+      "loss": 0.3233,
+      "step": 4246
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000272079701120797,
+      "loss": 0.5404,
+      "step": 4247
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002720672478206724,
+      "loss": 0.2921,
+      "step": 4248
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002720547945205479,
+      "loss": 0.247,
+      "step": 4249
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002720423412204234,
+      "loss": 0.2069,
+      "step": 4250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00027202988792029883,
+      "loss": 0.2658,
+      "step": 4251
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002720174346201743,
+      "loss": 0.2315,
+      "step": 4252
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00027200498132004975,
+      "loss": 0.6152,
+      "step": 4253
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00027199252801992524,
+      "loss": 0.2172,
+      "step": 4254
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002719800747198007,
+      "loss": 0.3024,
+      "step": 4255
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002719676214196762,
+      "loss": 0.2921,
+      "step": 4256
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00027195516811955164,
+      "loss": 0.247,
+      "step": 4257
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002719427148194271,
+      "loss": 0.254,
+      "step": 4258
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00027193026151930256,
+      "loss": 0.5145,
+      "step": 4259
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00027191780821917804,
+      "loss": 0.3757,
+      "step": 4260
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00027190535491905353,
+      "loss": 0.289,
+      "step": 4261
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000271892901618929,
+      "loss": 0.2101,
+      "step": 4262
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00027188044831880445,
+      "loss": 0.2628,
+      "step": 4263
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00027186799501867994,
+      "loss": 0.4238,
+      "step": 4264
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00027185554171855537,
+      "loss": 0.2819,
+      "step": 4265
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00027184308841843085,
+      "loss": 0.4131,
+      "step": 4266
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00027183063511830634,
+      "loss": 0.2711,
+      "step": 4267
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00027181818181818177,
+      "loss": 0.3031,
+      "step": 4268
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00027180572851805726,
+      "loss": 0.1809,
+      "step": 4269
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002717932752179327,
+      "loss": 0.2646,
+      "step": 4270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002717808219178082,
+      "loss": 0.3797,
+      "step": 4271
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00027176836861768366,
+      "loss": 0.4243,
+      "step": 4272
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00027175591531755915,
+      "loss": 0.2663,
+      "step": 4273
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002717434620174346,
+      "loss": 0.2727,
+      "step": 4274
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00027173100871731007,
+      "loss": 0.378,
+      "step": 4275
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002717185554171855,
+      "loss": 0.2342,
+      "step": 4276
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000271706102117061,
+      "loss": 0.2172,
+      "step": 4277
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00027169364881693647,
+      "loss": 0.4252,
+      "step": 4278
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00027168119551681196,
+      "loss": 0.257,
+      "step": 4279
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002716687422166874,
+      "loss": 0.3184,
+      "step": 4280
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002716562889165629,
+      "loss": 0.2498,
+      "step": 4281
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002716438356164383,
+      "loss": 0.4018,
+      "step": 4282
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002716313823163138,
+      "loss": 0.3116,
+      "step": 4283
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002716189290161893,
+      "loss": 0.4496,
+      "step": 4284
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002716064757160647,
+      "loss": 0.3298,
+      "step": 4285
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002715940224159402,
+      "loss": 0.2491,
+      "step": 4286
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00027158156911581563,
+      "loss": 0.2923,
+      "step": 4287
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002715691158156911,
+      "loss": 0.3078,
+      "step": 4288
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002715566625155666,
+      "loss": 0.3103,
+      "step": 4289
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002715442092154421,
+      "loss": 0.4223,
+      "step": 4290
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002715317559153175,
+      "loss": 0.6174,
+      "step": 4291
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000271519302615193,
+      "loss": 0.4003,
+      "step": 4292
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027150684931506844,
+      "loss": 0.2673,
+      "step": 4293
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027149439601494393,
+      "loss": 0.2831,
+      "step": 4294
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002714819427148194,
+      "loss": 0.3041,
+      "step": 4295
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002714694894146949,
+      "loss": 0.3985,
+      "step": 4296
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027145703611457033,
+      "loss": 0.4514,
+      "step": 4297
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027144458281444576,
+      "loss": 0.2648,
+      "step": 4298
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027143212951432125,
+      "loss": 0.4173,
+      "step": 4299
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027141967621419674,
+      "loss": 0.3374,
+      "step": 4300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002714072229140722,
+      "loss": 0.2996,
+      "step": 4301
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027139476961394765,
+      "loss": 0.2379,
+      "step": 4302
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027138231631382314,
+      "loss": 0.6197,
+      "step": 4303
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002713698630136986,
+      "loss": 0.2755,
+      "step": 4304
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027135740971357406,
+      "loss": 0.2824,
+      "step": 4305
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027134495641344955,
+      "loss": 0.232,
+      "step": 4306
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027133250311332503,
+      "loss": 0.3048,
+      "step": 4307
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027132004981320046,
+      "loss": 0.2111,
+      "step": 4308
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027130759651307595,
+      "loss": 0.3869,
+      "step": 4309
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002712951432129514,
+      "loss": 0.2879,
+      "step": 4310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027128268991282687,
+      "loss": 0.3949,
+      "step": 4311
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027127023661270236,
+      "loss": 0.2725,
+      "step": 4312
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00027125778331257784,
+      "loss": 0.4207,
+      "step": 4313
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002712453300124533,
+      "loss": 0.3696,
+      "step": 4314
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002712328767123287,
+      "loss": 0.3791,
+      "step": 4315
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002712204234122042,
+      "loss": 0.5939,
+      "step": 4316
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002712079701120797,
+      "loss": 0.2679,
+      "step": 4317
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00027119551681195516,
+      "loss": 0.2682,
+      "step": 4318
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002711830635118306,
+      "loss": 0.2374,
+      "step": 4319
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002711706102117061,
+      "loss": 0.2302,
+      "step": 4320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002711581569115815,
+      "loss": 0.1962,
+      "step": 4321
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000271145703611457,
+      "loss": 0.6935,
+      "step": 4322
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002711332503113325,
+      "loss": 0.2219,
+      "step": 4323
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000271120797011208,
+      "loss": 0.3901,
+      "step": 4324
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002711083437110834,
+      "loss": 0.2983,
+      "step": 4325
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002710958904109589,
+      "loss": 0.3083,
+      "step": 4326
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002710834371108343,
+      "loss": 0.3152,
+      "step": 4327
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002710709838107098,
+      "loss": 0.3965,
+      "step": 4328
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002710585305105853,
+      "loss": 0.3124,
+      "step": 4329
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00027104607721046073,
+      "loss": 0.2476,
+      "step": 4330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002710336239103362,
+      "loss": 0.2425,
+      "step": 4331
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00027102117061021165,
+      "loss": 0.4682,
+      "step": 4332
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00027100871731008713,
+      "loss": 0.2975,
+      "step": 4333
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002709962640099626,
+      "loss": 0.4044,
+      "step": 4334
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002709838107098381,
+      "loss": 0.2977,
+      "step": 4335
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00027097135740971354,
+      "loss": 0.3172,
+      "step": 4336
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000270958904109589,
+      "loss": 0.1877,
+      "step": 4337
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00027094645080946446,
+      "loss": 0.323,
+      "step": 4338
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00027093399750933994,
+      "loss": 0.2825,
+      "step": 4339
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00027092154420921543,
+      "loss": 0.3771,
+      "step": 4340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002709090909090909,
+      "loss": 0.3859,
+      "step": 4341
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00027089663760896635,
+      "loss": 0.2387,
+      "step": 4342
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00027088418430884183,
+      "loss": 0.2716,
+      "step": 4343
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027087173100871727,
+      "loss": 0.2959,
+      "step": 4344
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027085927770859275,
+      "loss": 0.1927,
+      "step": 4345
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027084682440846824,
+      "loss": 0.3256,
+      "step": 4346
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027083437110834367,
+      "loss": 0.6517,
+      "step": 4347
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027082191780821916,
+      "loss": 0.3831,
+      "step": 4348
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002708094645080946,
+      "loss": 0.238,
+      "step": 4349
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002707970112079701,
+      "loss": 0.294,
+      "step": 4350
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027078455790784556,
+      "loss": 0.6292,
+      "step": 4351
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027077210460772105,
+      "loss": 0.3052,
+      "step": 4352
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002707596513075965,
+      "loss": 0.6581,
+      "step": 4353
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027074719800747197,
+      "loss": 0.254,
+      "step": 4354
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002707347447073474,
+      "loss": 0.2548,
+      "step": 4355
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002707222914072229,
+      "loss": 0.2357,
+      "step": 4356
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027070983810709837,
+      "loss": 0.3392,
+      "step": 4357
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027069738480697386,
+      "loss": 0.2151,
+      "step": 4358
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002706849315068493,
+      "loss": 0.548,
+      "step": 4359
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002706724782067248,
+      "loss": 0.2876,
+      "step": 4360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002706600249066002,
+      "loss": 0.327,
+      "step": 4361
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002706475716064757,
+      "loss": 0.2326,
+      "step": 4362
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002706351183063512,
+      "loss": 0.3202,
+      "step": 4363
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002706226650062266,
+      "loss": 0.2515,
+      "step": 4364
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002706102117061021,
+      "loss": 0.448,
+      "step": 4365
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00027059775840597753,
+      "loss": 0.5945,
+      "step": 4366
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.000270585305105853,
+      "loss": 0.1902,
+      "step": 4367
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002705728518057285,
+      "loss": 0.2572,
+      "step": 4368
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.000270560398505604,
+      "loss": 0.3889,
+      "step": 4369
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002705479452054794,
+      "loss": 0.239,
+      "step": 4370
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002705354919053549,
+      "loss": 0.2192,
+      "step": 4371
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027052303860523034,
+      "loss": 0.499,
+      "step": 4372
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002705105853051058,
+      "loss": 0.2495,
+      "step": 4373
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002704981320049813,
+      "loss": 0.2922,
+      "step": 4374
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002704856787048568,
+      "loss": 0.1702,
+      "step": 4375
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027047322540473223,
+      "loss": 0.3529,
+      "step": 4376
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027046077210460766,
+      "loss": 0.2891,
+      "step": 4377
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027044831880448315,
+      "loss": 0.3838,
+      "step": 4378
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027043586550435863,
+      "loss": 0.6789,
+      "step": 4379
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002704234122042341,
+      "loss": 0.3213,
+      "step": 4380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027041095890410955,
+      "loss": 0.2372,
+      "step": 4381
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027039850560398504,
+      "loss": 0.261,
+      "step": 4382
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027038605230386047,
+      "loss": 0.2725,
+      "step": 4383
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027037359900373596,
+      "loss": 0.7255,
+      "step": 4384
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027036114570361144,
+      "loss": 0.2943,
+      "step": 4385
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027034869240348693,
+      "loss": 0.2288,
+      "step": 4386
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027033623910336236,
+      "loss": 0.1986,
+      "step": 4387
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027032378580323785,
+      "loss": 0.3081,
+      "step": 4388
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002703113325031133,
+      "loss": 0.3397,
+      "step": 4389
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027029887920298877,
+      "loss": 0.3501,
+      "step": 4390
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027028642590286425,
+      "loss": 0.3349,
+      "step": 4391
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027027397260273974,
+      "loss": 0.4398,
+      "step": 4392
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00027026151930261517,
+      "loss": 0.3171,
+      "step": 4393
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002702490660024906,
+      "loss": 0.2895,
+      "step": 4394
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002702366127023661,
+      "loss": 0.2339,
+      "step": 4395
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002702241594022416,
+      "loss": 0.34,
+      "step": 4396
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00027021170610211706,
+      "loss": 0.4701,
+      "step": 4397
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002701992528019925,
+      "loss": 0.3007,
+      "step": 4398
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.000270186799501868,
+      "loss": 0.4699,
+      "step": 4399
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002701743462017434,
+      "loss": 0.2239,
+      "step": 4400
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002701618929016189,
+      "loss": 0.3205,
+      "step": 4401
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002701494396014944,
+      "loss": 0.3394,
+      "step": 4402
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00027013698630136987,
+      "loss": 0.396,
+      "step": 4403
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002701245330012453,
+      "loss": 0.2334,
+      "step": 4404
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002701120797011208,
+      "loss": 0.245,
+      "step": 4405
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002700996264009962,
+      "loss": 0.2382,
+      "step": 4406
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002700871731008717,
+      "loss": 0.3347,
+      "step": 4407
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002700747198007472,
+      "loss": 0.2595,
+      "step": 4408
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002700622665006226,
+      "loss": 0.5556,
+      "step": 4409
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002700498132004981,
+      "loss": 0.3356,
+      "step": 4410
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00027003735990037354,
+      "loss": 0.4273,
+      "step": 4411
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00027002490660024903,
+      "loss": 0.2809,
+      "step": 4412
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002700124533001245,
+      "loss": 0.2059,
+      "step": 4413
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00027,
+      "loss": 0.4144,
+      "step": 4414
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00026998754669987544,
+      "loss": 0.3011,
+      "step": 4415
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002699750933997509,
+      "loss": 0.3927,
+      "step": 4416
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00026996264009962635,
+      "loss": 0.3013,
+      "step": 4417
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00026995018679950184,
+      "loss": 0.2737,
+      "step": 4418
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002699377334993773,
+      "loss": 0.2935,
+      "step": 4419
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002699252801992528,
+      "loss": 0.2189,
+      "step": 4420
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00026991282689912824,
+      "loss": 0.3271,
+      "step": 4421
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00026990037359900373,
+      "loss": 0.4546,
+      "step": 4422
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00026988792029887916,
+      "loss": 0.6029,
+      "step": 4423
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00026987546699875465,
+      "loss": 0.323,
+      "step": 4424
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00026986301369863014,
+      "loss": 0.2145,
+      "step": 4425
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00026985056039850557,
+      "loss": 0.2002,
+      "step": 4426
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00026983810709838105,
+      "loss": 0.4265,
+      "step": 4427
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002698256537982565,
+      "loss": 0.4327,
+      "step": 4428
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00026981320049813197,
+      "loss": 0.2682,
+      "step": 4429
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00026980074719800746,
+      "loss": 0.2981,
+      "step": 4430
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00026978829389788294,
+      "loss": 0.2565,
+      "step": 4431
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002697758405977584,
+      "loss": 0.2043,
+      "step": 4432
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00026976338729763386,
+      "loss": 0.2577,
+      "step": 4433
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002697509339975093,
+      "loss": 0.4319,
+      "step": 4434
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002697384806973848,
+      "loss": 0.2755,
+      "step": 4435
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00026972602739726027,
+      "loss": 0.2858,
+      "step": 4436
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00026971357409713575,
+      "loss": 0.2769,
+      "step": 4437
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002697011207970112,
+      "loss": 0.1977,
+      "step": 4438
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00026968866749688667,
+      "loss": 0.3674,
+      "step": 4439
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002696762141967621,
+      "loss": 0.4018,
+      "step": 4440
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002696637608966376,
+      "loss": 0.3097,
+      "step": 4441
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002696513075965131,
+      "loss": 0.3336,
+      "step": 4442
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002696388542963885,
+      "loss": 0.3248,
+      "step": 4443
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000269626400996264,
+      "loss": 0.2755,
+      "step": 4444
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00026961394769613943,
+      "loss": 0.1593,
+      "step": 4445
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002696014943960149,
+      "loss": 0.2387,
+      "step": 4446
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002695890410958904,
+      "loss": 0.4389,
+      "step": 4447
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002695765877957659,
+      "loss": 0.2614,
+      "step": 4448
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002695641344956413,
+      "loss": 0.2334,
+      "step": 4449
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002695516811955168,
+      "loss": 0.2603,
+      "step": 4450
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026953922789539224,
+      "loss": 0.2906,
+      "step": 4451
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002695267745952677,
+      "loss": 0.4017,
+      "step": 4452
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002695143212951432,
+      "loss": 0.5964,
+      "step": 4453
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002695018679950187,
+      "loss": 0.232,
+      "step": 4454
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026948941469489413,
+      "loss": 0.2375,
+      "step": 4455
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026947696139476956,
+      "loss": 0.2546,
+      "step": 4456
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026946450809464505,
+      "loss": 0.3849,
+      "step": 4457
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026945205479452053,
+      "loss": 0.2511,
+      "step": 4458
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.000269439601494396,
+      "loss": 0.6817,
+      "step": 4459
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026942714819427145,
+      "loss": 0.3053,
+      "step": 4460
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026941469489414694,
+      "loss": 0.295,
+      "step": 4461
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026940224159402237,
+      "loss": 0.2126,
+      "step": 4462
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026938978829389786,
+      "loss": 0.2678,
+      "step": 4463
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026937733499377334,
+      "loss": 0.3408,
+      "step": 4464
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026936488169364883,
+      "loss": 0.3067,
+      "step": 4465
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026935242839352426,
+      "loss": 0.3582,
+      "step": 4466
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026933997509339975,
+      "loss": 0.2908,
+      "step": 4467
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002693275217932752,
+      "loss": 0.4996,
+      "step": 4468
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026931506849315066,
+      "loss": 0.2557,
+      "step": 4469
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026930261519302615,
+      "loss": 0.3109,
+      "step": 4470
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002692901618929016,
+      "loss": 0.3178,
+      "step": 4471
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00026927770859277707,
+      "loss": 0.3987,
+      "step": 4472
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002692652552926525,
+      "loss": 0.2436,
+      "step": 4473
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.000269252801992528,
+      "loss": 0.2412,
+      "step": 4474
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002692403486924035,
+      "loss": 0.1816,
+      "step": 4475
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00026922789539227896,
+      "loss": 0.2259,
+      "step": 4476
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002692154420921544,
+      "loss": 0.3853,
+      "step": 4477
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002692029887920299,
+      "loss": 0.3587,
+      "step": 4478
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002691905354919053,
+      "loss": 0.3235,
+      "step": 4479
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002691780821917808,
+      "loss": 0.3214,
+      "step": 4480
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002691656288916563,
+      "loss": 0.2172,
+      "step": 4481
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00026915317559153177,
+      "loss": 0.3488,
+      "step": 4482
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002691407222914072,
+      "loss": 0.2365,
+      "step": 4483
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002691282689912827,
+      "loss": 0.5011,
+      "step": 4484
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002691158156911581,
+      "loss": 0.4289,
+      "step": 4485
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002691033623910336,
+      "loss": 0.2393,
+      "step": 4486
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002690909090909091,
+      "loss": 0.2661,
+      "step": 4487
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002690784557907845,
+      "loss": 0.231,
+      "step": 4488
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00026906600249066,
+      "loss": 0.2511,
+      "step": 4489
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00026905354919053544,
+      "loss": 0.3402,
+      "step": 4490
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00026904109589041093,
+      "loss": 0.46,
+      "step": 4491
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002690286425902864,
+      "loss": 0.4821,
+      "step": 4492
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002690161892901619,
+      "loss": 0.2844,
+      "step": 4493
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00026900373599003733,
+      "loss": 0.2926,
+      "step": 4494
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002689912826899128,
+      "loss": 0.3173,
+      "step": 4495
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00026897882938978825,
+      "loss": 0.3244,
+      "step": 4496
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00026896637608966374,
+      "loss": 0.4017,
+      "step": 4497
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002689539227895392,
+      "loss": 0.5429,
+      "step": 4498
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002689414694894147,
+      "loss": 0.3511,
+      "step": 4499
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00026892901618929014,
+      "loss": 0.262,
+      "step": 4500
+    },
+    {
+      "epoch": 1.72,
+      "eval_cer": 0.11505160270792282,
+      "eval_loss": 0.37477460503578186,
+      "eval_runtime": 46.989,
+      "eval_samples_per_second": 37.307,
+      "eval_steps_per_second": 4.682,
+      "eval_wer": 0.21036792324749967,
+      "step": 4500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00026891656288916563,
+      "loss": 0.3545,
+      "step": 4501
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00026890410958904106,
+      "loss": 0.3572,
+      "step": 4502
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00026889165628891655,
+      "loss": 0.567,
+      "step": 4503
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00026887920298879203,
+      "loss": 0.2402,
+      "step": 4504
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00026886674968866747,
+      "loss": 0.247,
+      "step": 4505
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00026885429638854295,
+      "loss": 0.2272,
+      "step": 4506
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002688418430884184,
+      "loss": 0.3283,
+      "step": 4507
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00026882938978829387,
+      "loss": 0.2207,
+      "step": 4508
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00026881693648816936,
+      "loss": 0.5004,
+      "step": 4509
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00026880448318804484,
+      "loss": 0.3349,
+      "step": 4510
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002687920298879203,
+      "loss": 0.2505,
+      "step": 4511
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00026877957658779576,
+      "loss": 0.209,
+      "step": 4512
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002687671232876712,
+      "loss": 0.2928,
+      "step": 4513
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002687546699875467,
+      "loss": 0.2555,
+      "step": 4514
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00026874221668742217,
+      "loss": 0.3617,
+      "step": 4515
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00026872976338729765,
+      "loss": 0.4393,
+      "step": 4516
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002687173100871731,
+      "loss": 0.3484,
+      "step": 4517
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002687048567870485,
+      "loss": 0.2144,
+      "step": 4518
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000268692403486924,
+      "loss": 0.2668,
+      "step": 4519
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002686799501867995,
+      "loss": 0.3231,
+      "step": 4520
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000268667496886675,
+      "loss": 0.3324,
+      "step": 4521
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002686550435865504,
+      "loss": 0.5727,
+      "step": 4522
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002686425902864259,
+      "loss": 0.2401,
+      "step": 4523
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002686301369863013,
+      "loss": 0.2062,
+      "step": 4524
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002686176836861768,
+      "loss": 0.2138,
+      "step": 4525
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002686052303860523,
+      "loss": 0.2451,
+      "step": 4526
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002685927770859278,
+      "loss": 0.3348,
+      "step": 4527
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002685803237858032,
+      "loss": 0.6208,
+      "step": 4528
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002685678704856787,
+      "loss": 0.352,
+      "step": 4529
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026855541718555413,
+      "loss": 0.2731,
+      "step": 4530
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002685429638854296,
+      "loss": 0.3053,
+      "step": 4531
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002685305105853051,
+      "loss": 0.2129,
+      "step": 4532
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026851805728518054,
+      "loss": 0.2235,
+      "step": 4533
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.000268505603985056,
+      "loss": 0.6665,
+      "step": 4534
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026849315068493146,
+      "loss": 0.225,
+      "step": 4535
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026848069738480694,
+      "loss": 0.2246,
+      "step": 4536
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026846824408468243,
+      "loss": 0.2286,
+      "step": 4537
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002684557907845579,
+      "loss": 0.2674,
+      "step": 4538
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026844333748443335,
+      "loss": 0.154,
+      "step": 4539
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026843088418430883,
+      "loss": 0.2925,
+      "step": 4540
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026841843088418427,
+      "loss": 0.4965,
+      "step": 4541
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026840597758405975,
+      "loss": 0.2415,
+      "step": 4542
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026839352428393524,
+      "loss": 0.2051,
+      "step": 4543
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026838107098381067,
+      "loss": 0.2664,
+      "step": 4544
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026836861768368616,
+      "loss": 0.501,
+      "step": 4545
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026835616438356164,
+      "loss": 0.322,
+      "step": 4546
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002683437110834371,
+      "loss": 0.4713,
+      "step": 4547
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026833125778331256,
+      "loss": 0.6263,
+      "step": 4548
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026831880448318805,
+      "loss": 0.3767,
+      "step": 4549
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002683063511830635,
+      "loss": 0.2943,
+      "step": 4550
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00026829389788293897,
+      "loss": 0.301,
+      "step": 4551
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002682814445828144,
+      "loss": 0.268,
+      "step": 4552
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002682689912826899,
+      "loss": 0.6506,
+      "step": 4553
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00026825653798256537,
+      "loss": 0.2381,
+      "step": 4554
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002682440846824408,
+      "loss": 0.3489,
+      "step": 4555
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002682316313823163,
+      "loss": 0.2429,
+      "step": 4556
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002682191780821918,
+      "loss": 0.2167,
+      "step": 4557
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002682067247820672,
+      "loss": 0.2481,
+      "step": 4558
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002681942714819427,
+      "loss": 0.6207,
+      "step": 4559
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002681818181818181,
+      "loss": 0.2566,
+      "step": 4560
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002681693648816936,
+      "loss": 0.2421,
+      "step": 4561
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002681569115815691,
+      "loss": 0.2561,
+      "step": 4562
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002681444582814446,
+      "loss": 0.2359,
+      "step": 4563
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00026813200498132,
+      "loss": 0.2714,
+      "step": 4564
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002681195516811955,
+      "loss": 0.352,
+      "step": 4565
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00026810709838107094,
+      "loss": 0.5184,
+      "step": 4566
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002680946450809464,
+      "loss": 0.2864,
+      "step": 4567
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002680821917808219,
+      "loss": 0.4486,
+      "step": 4568
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00026806973848069734,
+      "loss": 0.3478,
+      "step": 4569
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002680572851805728,
+      "loss": 0.2537,
+      "step": 4570
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00026804483188044826,
+      "loss": 0.3344,
+      "step": 4571
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00026803237858032374,
+      "loss": 0.4682,
+      "step": 4572
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00026801992528019923,
+      "loss": 0.3926,
+      "step": 4573
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002680074719800747,
+      "loss": 0.2513,
+      "step": 4574
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00026799501867995015,
+      "loss": 0.4381,
+      "step": 4575
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00026798256537982564,
+      "loss": 0.3735,
+      "step": 4576
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00026797011207970107,
+      "loss": 0.3406,
+      "step": 4577
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00026795765877957655,
+      "loss": 0.5338,
+      "step": 4578
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00026794520547945204,
+      "loss": 0.2457,
+      "step": 4579
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002679327521793275,
+      "loss": 0.3343,
+      "step": 4580
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00026792029887920296,
+      "loss": 0.2924,
+      "step": 4581
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002679078455790784,
+      "loss": 0.1868,
+      "step": 4582
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002678953922789539,
+      "loss": 0.423,
+      "step": 4583
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00026788293897882936,
+      "loss": 0.6506,
+      "step": 4584
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00026787048567870485,
+      "loss": 0.4058,
+      "step": 4585
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002678580323785803,
+      "loss": 0.25,
+      "step": 4586
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00026784557907845577,
+      "loss": 0.1989,
+      "step": 4587
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002678331257783312,
+      "loss": 0.2297,
+      "step": 4588
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002678206724782067,
+      "loss": 0.2123,
+      "step": 4589
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00026780821917808217,
+      "loss": 0.3721,
+      "step": 4590
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00026779576587795766,
+      "loss": 0.5655,
+      "step": 4591
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002677833125778331,
+      "loss": 0.2574,
+      "step": 4592
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002677708592777086,
+      "loss": 0.212,
+      "step": 4593
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.000267758405977584,
+      "loss": 0.4466,
+      "step": 4594
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002677459526774595,
+      "loss": 0.27,
+      "step": 4595
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.000267733499377335,
+      "loss": 0.2874,
+      "step": 4596
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002677210460772104,
+      "loss": 0.4502,
+      "step": 4597
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002677085927770859,
+      "loss": 0.4534,
+      "step": 4598
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00026769613947696133,
+      "loss": 0.2995,
+      "step": 4599
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002676836861768368,
+      "loss": 0.1854,
+      "step": 4600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002676712328767123,
+      "loss": 0.276,
+      "step": 4601
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002676587795765878,
+      "loss": 0.2542,
+      "step": 4602
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002676463262764632,
+      "loss": 0.5328,
+      "step": 4603
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002676338729763387,
+      "loss": 0.309,
+      "step": 4604
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026762141967621414,
+      "loss": 0.282,
+      "step": 4605
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026760896637608963,
+      "loss": 0.3644,
+      "step": 4606
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002675965130759651,
+      "loss": 0.2636,
+      "step": 4607
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002675840597758406,
+      "loss": 0.2029,
+      "step": 4608
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026757160647571603,
+      "loss": 0.3838,
+      "step": 4609
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002675591531755915,
+      "loss": 0.3192,
+      "step": 4610
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026754669987546695,
+      "loss": 0.2418,
+      "step": 4611
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026753424657534244,
+      "loss": 0.31,
+      "step": 4612
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002675217932752179,
+      "loss": 0.3942,
+      "step": 4613
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026750933997509336,
+      "loss": 0.2513,
+      "step": 4614
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026749688667496884,
+      "loss": 0.371,
+      "step": 4615
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002674844333748443,
+      "loss": 0.3946,
+      "step": 4616
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026747198007471976,
+      "loss": 0.3193,
+      "step": 4617
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026745952677459525,
+      "loss": 0.2506,
+      "step": 4618
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026744707347447073,
+      "loss": 0.3665,
+      "step": 4619
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026743462017434616,
+      "loss": 0.2367,
+      "step": 4620
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026742216687422165,
+      "loss": 0.2968,
+      "step": 4621
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002674097135740971,
+      "loss": 0.5763,
+      "step": 4622
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026739726027397257,
+      "loss": 0.2764,
+      "step": 4623
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026738480697384806,
+      "loss": 0.2641,
+      "step": 4624
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026737235367372354,
+      "loss": 0.394,
+      "step": 4625
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.000267359900373599,
+      "loss": 0.2006,
+      "step": 4626
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026734744707347446,
+      "loss": 0.3494,
+      "step": 4627
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002673349937733499,
+      "loss": 0.4603,
+      "step": 4628
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002673225404732254,
+      "loss": 0.3464,
+      "step": 4629
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00026731008717310086,
+      "loss": 0.2516,
+      "step": 4630
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002672976338729763,
+      "loss": 0.2138,
+      "step": 4631
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002672851805728518,
+      "loss": 0.1754,
+      "step": 4632
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002672727272727272,
+      "loss": 0.3603,
+      "step": 4633
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002672602739726027,
+      "loss": 0.3671,
+      "step": 4634
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002672478206724782,
+      "loss": 0.2758,
+      "step": 4635
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002672353673723537,
+      "loss": 0.2446,
+      "step": 4636
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002672229140722291,
+      "loss": 0.4261,
+      "step": 4637
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002672104607721046,
+      "loss": 0.2935,
+      "step": 4638
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00026719800747198,
+      "loss": 0.3618,
+      "step": 4639
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002671855541718555,
+      "loss": 0.503,
+      "step": 4640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.000267173100871731,
+      "loss": 0.3806,
+      "step": 4641
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002671606475716065,
+      "loss": 0.2336,
+      "step": 4642
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002671481942714819,
+      "loss": 0.2378,
+      "step": 4643
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00026713574097135735,
+      "loss": 0.301,
+      "step": 4644
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00026712328767123283,
+      "loss": 0.2599,
+      "step": 4645
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002671108343711083,
+      "loss": 0.3994,
+      "step": 4646
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002670983810709838,
+      "loss": 0.4411,
+      "step": 4647
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00026708592777085924,
+      "loss": 0.2169,
+      "step": 4648
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002670734744707347,
+      "loss": 0.2775,
+      "step": 4649
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00026706102117061016,
+      "loss": 0.2592,
+      "step": 4650
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00026704856787048564,
+      "loss": 0.2808,
+      "step": 4651
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00026703611457036113,
+      "loss": 0.378,
+      "step": 4652
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002670236612702366,
+      "loss": 0.5269,
+      "step": 4653
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00026701120797011205,
+      "loss": 0.2952,
+      "step": 4654
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00026699875466998753,
+      "loss": 0.2842,
+      "step": 4655
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00026698630136986297,
+      "loss": 0.2322,
+      "step": 4656
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00026697384806973845,
+      "loss": 0.2047,
+      "step": 4657
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00026696139476961394,
+      "loss": 0.2537,
+      "step": 4658
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00026694894146948937,
+      "loss": 0.5588,
+      "step": 4659
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00026693648816936486,
+      "loss": 0.4336,
+      "step": 4660
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002669240348692403,
+      "loss": 0.2223,
+      "step": 4661
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002669115815691158,
+      "loss": 0.3312,
+      "step": 4662
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00026689912826899126,
+      "loss": 0.2313,
+      "step": 4663
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00026688667496886675,
+      "loss": 0.1999,
+      "step": 4664
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002668742216687422,
+      "loss": 0.3812,
+      "step": 4665
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00026686176836861767,
+      "loss": 0.4925,
+      "step": 4666
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002668493150684931,
+      "loss": 0.2671,
+      "step": 4667
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002668368617683686,
+      "loss": 0.187,
+      "step": 4668
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00026682440846824407,
+      "loss": 0.2108,
+      "step": 4669
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00026681195516811956,
+      "loss": 0.2493,
+      "step": 4670
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000266799501867995,
+      "loss": 0.3187,
+      "step": 4671
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002667870485678705,
+      "loss": 0.4343,
+      "step": 4672
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002667745952677459,
+      "loss": 0.2027,
+      "step": 4673
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002667621419676214,
+      "loss": 0.244,
+      "step": 4674
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002667496886674969,
+      "loss": 0.3558,
+      "step": 4675
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002667372353673723,
+      "loss": 0.2541,
+      "step": 4676
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002667247820672478,
+      "loss": 0.2497,
+      "step": 4677
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00026671232876712323,
+      "loss": 0.4717,
+      "step": 4678
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002666998754669987,
+      "loss": 0.2025,
+      "step": 4679
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002666874221668742,
+      "loss": 0.3478,
+      "step": 4680
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002666749688667497,
+      "loss": 0.3208,
+      "step": 4681
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002666625155666251,
+      "loss": 0.2227,
+      "step": 4682
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002666500622665006,
+      "loss": 0.2758,
+      "step": 4683
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026663760896637604,
+      "loss": 0.4894,
+      "step": 4684
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002666251556662515,
+      "loss": 0.2521,
+      "step": 4685
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000266612702366127,
+      "loss": 0.2303,
+      "step": 4686
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002666002490660025,
+      "loss": 0.3442,
+      "step": 4687
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026658779576587793,
+      "loss": 0.2575,
+      "step": 4688
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002665753424657534,
+      "loss": 0.2848,
+      "step": 4689
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026656288916562885,
+      "loss": 0.3775,
+      "step": 4690
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026655043586550433,
+      "loss": 0.3675,
+      "step": 4691
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002665379825653798,
+      "loss": 0.2531,
+      "step": 4692
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026652552926525525,
+      "loss": 0.2757,
+      "step": 4693
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026651307596513074,
+      "loss": 0.2599,
+      "step": 4694
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026650062266500617,
+      "loss": 0.3278,
+      "step": 4695
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026648816936488166,
+      "loss": 0.3839,
+      "step": 4696
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026647571606475714,
+      "loss": 0.4119,
+      "step": 4697
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026646326276463263,
+      "loss": 0.2293,
+      "step": 4698
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026645080946450806,
+      "loss": 0.2555,
+      "step": 4699
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026643835616438355,
+      "loss": 0.304,
+      "step": 4700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000266425902864259,
+      "loss": 0.2506,
+      "step": 4701
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026641344956413447,
+      "loss": 0.6092,
+      "step": 4702
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026640099626400995,
+      "loss": 0.3651,
+      "step": 4703
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026638854296388544,
+      "loss": 0.2472,
+      "step": 4704
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026637608966376087,
+      "loss": 0.2402,
+      "step": 4705
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026636363636363636,
+      "loss": 0.3067,
+      "step": 4706
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002663511830635118,
+      "loss": 0.2917,
+      "step": 4707
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002663387297633873,
+      "loss": 0.4519,
+      "step": 4708
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00026632627646326276,
+      "loss": 0.285,
+      "step": 4709
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002663138231631382,
+      "loss": 0.4895,
+      "step": 4710
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002663013698630137,
+      "loss": 0.2431,
+      "step": 4711
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002662889165628891,
+      "loss": 0.2679,
+      "step": 4712
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002662764632627646,
+      "loss": 0.2103,
+      "step": 4713
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002662640099626401,
+      "loss": 0.2794,
+      "step": 4714
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00026625155666251557,
+      "loss": 0.4632,
+      "step": 4715
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000266239103362391,
+      "loss": 0.3624,
+      "step": 4716
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002662266500622665,
+      "loss": 0.3623,
+      "step": 4717
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002662141967621419,
+      "loss": 0.2321,
+      "step": 4718
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002662017434620174,
+      "loss": 0.2366,
+      "step": 4719
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002661892901618929,
+      "loss": 0.2628,
+      "step": 4720
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002661768368617684,
+      "loss": 0.2739,
+      "step": 4721
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002661643835616438,
+      "loss": 0.4579,
+      "step": 4722
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00026615193026151924,
+      "loss": 0.2071,
+      "step": 4723
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00026613947696139473,
+      "loss": 0.2226,
+      "step": 4724
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002661270236612702,
+      "loss": 0.2481,
+      "step": 4725
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002661145703611457,
+      "loss": 0.2874,
+      "step": 4726
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00026610211706102114,
+      "loss": 0.1959,
+      "step": 4727
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002660896637608966,
+      "loss": 0.3698,
+      "step": 4728
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00026607721046077205,
+      "loss": 0.3397,
+      "step": 4729
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00026606475716064754,
+      "loss": 0.2151,
+      "step": 4730
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000266052303860523,
+      "loss": 0.3391,
+      "step": 4731
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002660398505603985,
+      "loss": 0.2313,
+      "step": 4732
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00026602739726027394,
+      "loss": 0.3725,
+      "step": 4733
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00026601494396014943,
+      "loss": 0.7478,
+      "step": 4734
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00026600249066002486,
+      "loss": 0.3246,
+      "step": 4735
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00026599003735990035,
+      "loss": 0.3026,
+      "step": 4736
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00026597758405977584,
+      "loss": 0.2325,
+      "step": 4737
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00026596513075965127,
+      "loss": 0.2259,
+      "step": 4738
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00026595267745952675,
+      "loss": 0.3181,
+      "step": 4739
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002659402241594022,
+      "loss": 0.2906,
+      "step": 4740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00026592777085927767,
+      "loss": 0.4183,
+      "step": 4741
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00026591531755915316,
+      "loss": 0.2793,
+      "step": 4742
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00026590286425902864,
+      "loss": 0.2217,
+      "step": 4743
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002658904109589041,
+      "loss": 0.2924,
+      "step": 4744
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00026587795765877956,
+      "loss": 0.305,
+      "step": 4745
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000265865504358655,
+      "loss": 0.399,
+      "step": 4746
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002658530510585305,
+      "loss": 0.526,
+      "step": 4747
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00026584059775840597,
+      "loss": 0.395,
+      "step": 4748
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00026582814445828145,
+      "loss": 0.2814,
+      "step": 4749
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002658156911581569,
+      "loss": 0.237,
+      "step": 4750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00026580323785803237,
+      "loss": 0.1884,
+      "step": 4751
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002657907845579078,
+      "loss": 0.3269,
+      "step": 4752
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002657783312577833,
+      "loss": 0.404,
+      "step": 4753
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002657658779576588,
+      "loss": 0.2495,
+      "step": 4754
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002657534246575342,
+      "loss": 0.2382,
+      "step": 4755
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002657409713574097,
+      "loss": 0.2079,
+      "step": 4756
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00026572851805728513,
+      "loss": 0.273,
+      "step": 4757
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002657160647571606,
+      "loss": 0.2249,
+      "step": 4758
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002657036114570361,
+      "loss": 0.558,
+      "step": 4759
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002656911581569116,
+      "loss": 0.242,
+      "step": 4760
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000265678704856787,
+      "loss": 0.7534,
+      "step": 4761
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002656662515566625,
+      "loss": 0.3791,
+      "step": 4762
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026565379825653794,
+      "loss": 0.2758,
+      "step": 4763
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002656413449564134,
+      "loss": 0.3268,
+      "step": 4764
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002656288916562889,
+      "loss": 0.3651,
+      "step": 4765
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002656164383561644,
+      "loss": 0.5648,
+      "step": 4766
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026560398505603983,
+      "loss": 0.2725,
+      "step": 4767
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002655915317559153,
+      "loss": 0.333,
+      "step": 4768
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026557907845579075,
+      "loss": 0.3451,
+      "step": 4769
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026556662515566623,
+      "loss": 0.2807,
+      "step": 4770
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002655541718555417,
+      "loss": 0.3407,
+      "step": 4771
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026554171855541715,
+      "loss": 0.6241,
+      "step": 4772
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026552926525529264,
+      "loss": 0.3498,
+      "step": 4773
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026551681195516807,
+      "loss": 0.323,
+      "step": 4774
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026550435865504356,
+      "loss": 0.2836,
+      "step": 4775
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026549190535491904,
+      "loss": 0.2448,
+      "step": 4776
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026547945205479453,
+      "loss": 0.4573,
+      "step": 4777
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026546699875466996,
+      "loss": 0.4244,
+      "step": 4778
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026545454545454545,
+      "loss": 0.2231,
+      "step": 4779
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002654420921544209,
+      "loss": 0.2473,
+      "step": 4780
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026542963885429636,
+      "loss": 0.331,
+      "step": 4781
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026541718555417185,
+      "loss": 0.2459,
+      "step": 4782
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026540473225404734,
+      "loss": 0.3321,
+      "step": 4783
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00026539227895392277,
+      "loss": 0.8426,
+      "step": 4784
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002653798256537982,
+      "loss": 0.2907,
+      "step": 4785
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002653673723536737,
+      "loss": 0.235,
+      "step": 4786
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002653549190535492,
+      "loss": 0.3395,
+      "step": 4787
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00026534246575342466,
+      "loss": 0.3108,
+      "step": 4788
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002653300124533001,
+      "loss": 0.3254,
+      "step": 4789
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002653175591531756,
+      "loss": 0.4513,
+      "step": 4790
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000265305105853051,
+      "loss": 0.4315,
+      "step": 4791
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002652926525529265,
+      "loss": 0.3562,
+      "step": 4792
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000265280199252802,
+      "loss": 0.2697,
+      "step": 4793
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00026526774595267747,
+      "loss": 0.2003,
+      "step": 4794
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002652552926525529,
+      "loss": 0.2706,
+      "step": 4795
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002652428393524284,
+      "loss": 0.2355,
+      "step": 4796
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002652303860523038,
+      "loss": 0.36,
+      "step": 4797
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002652179327521793,
+      "loss": 0.3728,
+      "step": 4798
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002652054794520548,
+      "loss": 0.2745,
+      "step": 4799
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002651930261519303,
+      "loss": 0.2444,
+      "step": 4800
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002651805728518057,
+      "loss": 0.308,
+      "step": 4801
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00026516811955168114,
+      "loss": 0.3826,
+      "step": 4802
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00026515566625155663,
+      "loss": 0.4276,
+      "step": 4803
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002651432129514321,
+      "loss": 0.3214,
+      "step": 4804
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002651307596513076,
+      "loss": 0.2641,
+      "step": 4805
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00026511830635118303,
+      "loss": 0.2532,
+      "step": 4806
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002651058530510585,
+      "loss": 0.2411,
+      "step": 4807
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00026509339975093395,
+      "loss": 0.3025,
+      "step": 4808
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00026508094645080944,
+      "loss": 0.536,
+      "step": 4809
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002650684931506849,
+      "loss": 0.3236,
+      "step": 4810
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002650560398505604,
+      "loss": 0.3465,
+      "step": 4811
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00026504358655043584,
+      "loss": 0.2927,
+      "step": 4812
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00026503113325031133,
+      "loss": 0.3028,
+      "step": 4813
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00026501867995018676,
+      "loss": 0.5168,
+      "step": 4814
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00026500622665006225,
+      "loss": 0.4865,
+      "step": 4815
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00026499377334993773,
+      "loss": 0.4624,
+      "step": 4816
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00026498132004981317,
+      "loss": 0.3312,
+      "step": 4817
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00026496886674968865,
+      "loss": 0.2863,
+      "step": 4818
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002649564134495641,
+      "loss": 0.2928,
+      "step": 4819
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00026494396014943957,
+      "loss": 0.2245,
+      "step": 4820
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00026493150684931506,
+      "loss": 0.2422,
+      "step": 4821
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00026491905354919054,
+      "loss": 0.4462,
+      "step": 4822
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000264906600249066,
+      "loss": 0.2668,
+      "step": 4823
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00026489414694894146,
+      "loss": 0.3185,
+      "step": 4824
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002648816936488169,
+      "loss": 0.2199,
+      "step": 4825
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002648692403486924,
+      "loss": 0.2957,
+      "step": 4826
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00026485678704856787,
+      "loss": 0.2904,
+      "step": 4827
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00026484433374844335,
+      "loss": 0.6255,
+      "step": 4828
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002648318804483188,
+      "loss": 0.2775,
+      "step": 4829
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00026481942714819427,
+      "loss": 0.3077,
+      "step": 4830
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002648069738480697,
+      "loss": 0.2867,
+      "step": 4831
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002647945205479452,
+      "loss": 0.2938,
+      "step": 4832
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002647820672478207,
+      "loss": 0.2606,
+      "step": 4833
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002647696139476961,
+      "loss": 0.4872,
+      "step": 4834
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002647571606475716,
+      "loss": 0.389,
+      "step": 4835
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000264744707347447,
+      "loss": 0.277,
+      "step": 4836
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002647322540473225,
+      "loss": 0.2802,
+      "step": 4837
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000264719800747198,
+      "loss": 0.2069,
+      "step": 4838
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002647073474470735,
+      "loss": 0.3062,
+      "step": 4839
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002646948941469489,
+      "loss": 0.2693,
+      "step": 4840
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002646824408468244,
+      "loss": 0.3834,
+      "step": 4841
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026466998754669983,
+      "loss": 0.3001,
+      "step": 4842
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002646575342465753,
+      "loss": 0.2938,
+      "step": 4843
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002646450809464508,
+      "loss": 0.2233,
+      "step": 4844
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002646326276463263,
+      "loss": 0.1865,
+      "step": 4845
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002646201743462017,
+      "loss": 0.2796,
+      "step": 4846
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002646077210460772,
+      "loss": 0.3136,
+      "step": 4847
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026459526774595264,
+      "loss": 0.5066,
+      "step": 4848
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026458281444582813,
+      "loss": 0.2654,
+      "step": 4849
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002645703611457036,
+      "loss": 0.3189,
+      "step": 4850
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026455790784557905,
+      "loss": 0.2687,
+      "step": 4851
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026454545454545453,
+      "loss": 0.4288,
+      "step": 4852
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026453300124532997,
+      "loss": 0.4996,
+      "step": 4853
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026452054794520545,
+      "loss": 0.409,
+      "step": 4854
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026450809464508094,
+      "loss": 0.2712,
+      "step": 4855
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002644956413449564,
+      "loss": 0.2977,
+      "step": 4856
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026448318804483186,
+      "loss": 0.206,
+      "step": 4857
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026447073474470734,
+      "loss": 0.2378,
+      "step": 4858
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002644582814445828,
+      "loss": 0.4779,
+      "step": 4859
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026444582814445826,
+      "loss": 0.3162,
+      "step": 4860
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026443337484433375,
+      "loss": 0.2423,
+      "step": 4861
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026442092154420923,
+      "loss": 0.3134,
+      "step": 4862
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00026440846824408467,
+      "loss": 0.4651,
+      "step": 4863
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002643960149439601,
+      "loss": 0.3287,
+      "step": 4864
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002643835616438356,
+      "loss": 0.4339,
+      "step": 4865
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00026437110834371107,
+      "loss": 0.3435,
+      "step": 4866
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00026435865504358656,
+      "loss": 0.2862,
+      "step": 4867
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000264346201743462,
+      "loss": 0.2592,
+      "step": 4868
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002643337484433375,
+      "loss": 0.2164,
+      "step": 4869
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002643212951432129,
+      "loss": 0.3074,
+      "step": 4870
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002643088418430884,
+      "loss": 0.1912,
+      "step": 4871
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002642963885429639,
+      "loss": 0.475,
+      "step": 4872
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002642839352428393,
+      "loss": 0.2452,
+      "step": 4873
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002642714819427148,
+      "loss": 0.2012,
+      "step": 4874
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002642590286425903,
+      "loss": 0.4244,
+      "step": 4875
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002642465753424657,
+      "loss": 0.248,
+      "step": 4876
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002642341220423412,
+      "loss": 0.352,
+      "step": 4877
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002642216687422167,
+      "loss": 0.4269,
+      "step": 4878
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002642092154420921,
+      "loss": 0.274,
+      "step": 4879
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002641967621419676,
+      "loss": 0.2842,
+      "step": 4880
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00026418430884184304,
+      "loss": 0.2033,
+      "step": 4881
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002641718555417185,
+      "loss": 0.2637,
+      "step": 4882
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000264159402241594,
+      "loss": 0.2226,
+      "step": 4883
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00026414694894146944,
+      "loss": 0.5553,
+      "step": 4884
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00026413449564134493,
+      "loss": 0.2398,
+      "step": 4885
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002641220423412204,
+      "loss": 0.2784,
+      "step": 4886
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00026410958904109585,
+      "loss": 0.4546,
+      "step": 4887
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00026409713574097134,
+      "loss": 0.6712,
+      "step": 4888
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002640846824408468,
+      "loss": 0.3406,
+      "step": 4889
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00026407222914072225,
+      "loss": 0.3029,
+      "step": 4890
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00026405977584059774,
+      "loss": 0.4358,
+      "step": 4891
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002640473225404732,
+      "loss": 0.2865,
+      "step": 4892
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00026403486924034866,
+      "loss": 0.2595,
+      "step": 4893
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00026402241594022414,
+      "loss": 0.2692,
+      "step": 4894
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002640099626400996,
+      "loss": 0.2277,
+      "step": 4895
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00026399750933997506,
+      "loss": 0.2535,
+      "step": 4896
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00026398505603985055,
+      "loss": 0.4018,
+      "step": 4897
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000263972602739726,
+      "loss": 0.2641,
+      "step": 4898
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00026396014943960147,
+      "loss": 0.2643,
+      "step": 4899
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00026394769613947695,
+      "loss": 0.2727,
+      "step": 4900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002639352428393524,
+      "loss": 0.2576,
+      "step": 4901
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00026392278953922787,
+      "loss": 0.3165,
+      "step": 4902
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00026391033623910336,
+      "loss": 0.5739,
+      "step": 4903
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002638978829389788,
+      "loss": 0.2503,
+      "step": 4904
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002638854296388543,
+      "loss": 0.1957,
+      "step": 4905
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002638729763387297,
+      "loss": 0.2851,
+      "step": 4906
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002638605230386052,
+      "loss": 0.1974,
+      "step": 4907
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002638480697384807,
+      "loss": 0.512,
+      "step": 4908
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00026383561643835617,
+      "loss": 0.3602,
+      "step": 4909
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002638231631382316,
+      "loss": 0.266,
+      "step": 4910
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002638107098381071,
+      "loss": 0.3782,
+      "step": 4911
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002637982565379825,
+      "loss": 0.2154,
+      "step": 4912
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000263785803237858,
+      "loss": 0.3298,
+      "step": 4913
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002637733499377335,
+      "loss": 0.3557,
+      "step": 4914
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002637608966376089,
+      "loss": 0.2957,
+      "step": 4915
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002637484433374844,
+      "loss": 0.4689,
+      "step": 4916
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00026373599003735984,
+      "loss": 0.3448,
+      "step": 4917
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026372353673723533,
+      "loss": 0.2704,
+      "step": 4918
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002637110834371108,
+      "loss": 0.2948,
+      "step": 4919
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002636986301369863,
+      "loss": 0.2466,
+      "step": 4920
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026368617683686173,
+      "loss": 0.2724,
+      "step": 4921
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002636737235367372,
+      "loss": 0.4364,
+      "step": 4922
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026366127023661265,
+      "loss": 0.3001,
+      "step": 4923
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026364881693648814,
+      "loss": 0.2819,
+      "step": 4924
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002636363636363636,
+      "loss": 0.354,
+      "step": 4925
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026362391033623906,
+      "loss": 0.2436,
+      "step": 4926
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026361145703611454,
+      "loss": 0.2384,
+      "step": 4927
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026359900373599,
+      "loss": 0.5981,
+      "step": 4928
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026358655043586546,
+      "loss": 0.3145,
+      "step": 4929
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026357409713574095,
+      "loss": 0.2591,
+      "step": 4930
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026356164383561643,
+      "loss": 0.2379,
+      "step": 4931
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026354919053549186,
+      "loss": 0.3026,
+      "step": 4932
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026353673723536735,
+      "loss": 0.2907,
+      "step": 4933
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002635242839352428,
+      "loss": 0.4716,
+      "step": 4934
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026351183063511827,
+      "loss": 0.4653,
+      "step": 4935
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026349937733499376,
+      "loss": 0.2051,
+      "step": 4936
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026348692403486924,
+      "loss": 0.2293,
+      "step": 4937
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002634744707347447,
+      "loss": 0.2548,
+      "step": 4938
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026346201743462016,
+      "loss": 0.2028,
+      "step": 4939
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002634495641344956,
+      "loss": 0.3666,
+      "step": 4940
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002634371108343711,
+      "loss": 0.2818,
+      "step": 4941
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00026342465753424656,
+      "loss": 0.2819,
+      "step": 4942
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.000263412204234122,
+      "loss": 0.3649,
+      "step": 4943
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002633997509339975,
+      "loss": 0.4053,
+      "step": 4944
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002633872976338729,
+      "loss": 0.2157,
+      "step": 4945
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002633748443337484,
+      "loss": 0.3869,
+      "step": 4946
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002633623910336239,
+      "loss": 0.4397,
+      "step": 4947
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002633499377334994,
+      "loss": 0.381,
+      "step": 4948
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002633374844333748,
+      "loss": 0.2934,
+      "step": 4949
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002633250311332503,
+      "loss": 0.2545,
+      "step": 4950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002633125778331257,
+      "loss": 0.3183,
+      "step": 4951
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002633001245330012,
+      "loss": 0.2573,
+      "step": 4952
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002632876712328767,
+      "loss": 0.3396,
+      "step": 4953
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002632752179327522,
+      "loss": 0.2609,
+      "step": 4954
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002632627646326276,
+      "loss": 0.2202,
+      "step": 4955
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002632503113325031,
+      "loss": 0.3627,
+      "step": 4956
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00026323785803237853,
+      "loss": 0.2611,
+      "step": 4957
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000263225404732254,
+      "loss": 0.3115,
+      "step": 4958
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002632129514321295,
+      "loss": 0.4695,
+      "step": 4959
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00026320049813200494,
+      "loss": 0.2935,
+      "step": 4960
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002631880448318804,
+      "loss": 0.4066,
+      "step": 4961
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00026317559153175586,
+      "loss": 0.2054,
+      "step": 4962
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00026316313823163134,
+      "loss": 0.3835,
+      "step": 4963
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00026315068493150683,
+      "loss": 0.1926,
+      "step": 4964
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002631382316313823,
+      "loss": 0.3249,
+      "step": 4965
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00026312577833125775,
+      "loss": 0.3709,
+      "step": 4966
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00026311332503113323,
+      "loss": 0.3551,
+      "step": 4967
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00026310087173100867,
+      "loss": 0.1983,
+      "step": 4968
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00026308841843088415,
+      "loss": 0.2456,
+      "step": 4969
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00026307596513075964,
+      "loss": 0.3732,
+      "step": 4970
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002630635118306351,
+      "loss": 0.2375,
+      "step": 4971
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00026305105853051056,
+      "loss": 0.489,
+      "step": 4972
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00026303860523038604,
+      "loss": 0.2596,
+      "step": 4973
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002630261519302615,
+      "loss": 0.819,
+      "step": 4974
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00026301369863013696,
+      "loss": 0.3966,
+      "step": 4975
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00026300124533001245,
+      "loss": 0.22,
+      "step": 4976
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002629887920298879,
+      "loss": 0.3945,
+      "step": 4977
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00026297633872976337,
+      "loss": 0.4752,
+      "step": 4978
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002629638854296388,
+      "loss": 0.2933,
+      "step": 4979
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002629514321295143,
+      "loss": 0.2791,
+      "step": 4980
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00026293897882938977,
+      "loss": 0.2695,
+      "step": 4981
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00026292652552926526,
+      "loss": 0.2924,
+      "step": 4982
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002629140722291407,
+      "loss": 0.3236,
+      "step": 4983
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002629016189290162,
+      "loss": 0.4123,
+      "step": 4984
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002628891656288916,
+      "loss": 0.2664,
+      "step": 4985
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002628767123287671,
+      "loss": 0.2492,
+      "step": 4986
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002628642590286426,
+      "loss": 0.3947,
+      "step": 4987
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00026285180572851807,
+      "loss": 0.3523,
+      "step": 4988
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002628393524283935,
+      "loss": 0.3075,
+      "step": 4989
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00026282689912826893,
+      "loss": 0.3939,
+      "step": 4990
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002628144458281444,
+      "loss": 0.4748,
+      "step": 4991
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002628019925280199,
+      "loss": 0.2636,
+      "step": 4992
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002627895392278954,
+      "loss": 0.254,
+      "step": 4993
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002627770859277708,
+      "loss": 0.2806,
+      "step": 4994
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002627646326276463,
+      "loss": 0.3361,
+      "step": 4995
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00026275217932752174,
+      "loss": 0.3671,
+      "step": 4996
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002627397260273972,
+      "loss": 0.3635,
+      "step": 4997
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002627272727272727,
+      "loss": 0.331,
+      "step": 4998
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002627148194271482,
+      "loss": 0.301,
+      "step": 4999
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026270236612702363,
+      "loss": 0.3138,
+      "step": 5000
+    },
+    {
+      "epoch": 1.92,
+      "eval_cer": 0.11336597928087871,
+      "eval_loss": 0.3825168311595917,
+      "eval_runtime": 46.6598,
+      "eval_samples_per_second": 37.57,
+      "eval_steps_per_second": 4.715,
+      "eval_wer": 0.19934832782730688,
+      "step": 5000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002626899128268991,
+      "loss": 0.279,
+      "step": 5001
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026267745952677455,
+      "loss": 0.3618,
+      "step": 5002
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026266500622665003,
+      "loss": 0.3572,
+      "step": 5003
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002626525529265255,
+      "loss": 0.2878,
+      "step": 5004
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026264009962640095,
+      "loss": 0.4394,
+      "step": 5005
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026262764632627644,
+      "loss": 0.322,
+      "step": 5006
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026261519302615187,
+      "loss": 0.315,
+      "step": 5007
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026260273972602736,
+      "loss": 0.3433,
+      "step": 5008
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026259028642590284,
+      "loss": 0.4713,
+      "step": 5009
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026257783312577833,
+      "loss": 0.2611,
+      "step": 5010
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026256537982565376,
+      "loss": 0.2409,
+      "step": 5011
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026255292652552925,
+      "loss": 0.253,
+      "step": 5012
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002625404732254047,
+      "loss": 0.3369,
+      "step": 5013
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026252801992528017,
+      "loss": 0.2653,
+      "step": 5014
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026251556662515565,
+      "loss": 0.2794,
+      "step": 5015
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026250311332503114,
+      "loss": 0.3104,
+      "step": 5016
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026249066002490657,
+      "loss": 0.4416,
+      "step": 5017
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026247820672478206,
+      "loss": 0.3211,
+      "step": 5018
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002624657534246575,
+      "loss": 0.3681,
+      "step": 5019
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.000262453300124533,
+      "loss": 0.2116,
+      "step": 5020
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00026244084682440846,
+      "loss": 0.2629,
+      "step": 5021
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002624283935242839,
+      "loss": 0.5396,
+      "step": 5022
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002624159402241594,
+      "loss": 0.3252,
+      "step": 5023
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002624034869240348,
+      "loss": 0.2211,
+      "step": 5024
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002623910336239103,
+      "loss": 0.2453,
+      "step": 5025
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002623785803237858,
+      "loss": 0.2796,
+      "step": 5026
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00026236612702366127,
+      "loss": 0.4468,
+      "step": 5027
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002623536737235367,
+      "loss": 0.4574,
+      "step": 5028
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002623412204234122,
+      "loss": 0.2513,
+      "step": 5029
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002623287671232876,
+      "loss": 0.2845,
+      "step": 5030
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002623163138231631,
+      "loss": 0.214,
+      "step": 5031
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002623038605230386,
+      "loss": 0.2418,
+      "step": 5032
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002622914072229141,
+      "loss": 0.2814,
+      "step": 5033
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002622789539227895,
+      "loss": 0.4229,
+      "step": 5034
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.000262266500622665,
+      "loss": 0.2376,
+      "step": 5035
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00026225404732254043,
+      "loss": 0.2097,
+      "step": 5036
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002622415940224159,
+      "loss": 0.2519,
+      "step": 5037
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002622291407222914,
+      "loss": 0.2094,
+      "step": 5038
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00026221668742216684,
+      "loss": 0.2011,
+      "step": 5039
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002622042341220423,
+      "loss": 0.2375,
+      "step": 5040
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00026219178082191775,
+      "loss": 0.684,
+      "step": 5041
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00026217932752179324,
+      "loss": 0.5262,
+      "step": 5042
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002621668742216687,
+      "loss": 0.3804,
+      "step": 5043
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002621544209215442,
+      "loss": 0.2394,
+      "step": 5044
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00026214196762141964,
+      "loss": 0.2629,
+      "step": 5045
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00026212951432129513,
+      "loss": 0.1968,
+      "step": 5046
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00026211706102117056,
+      "loss": 0.6688,
+      "step": 5047
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00026210460772104605,
+      "loss": 0.2567,
+      "step": 5048
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00026209215442092154,
+      "loss": 0.2909,
+      "step": 5049
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000262079701120797,
+      "loss": 0.3137,
+      "step": 5050
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00026206724782067245,
+      "loss": 0.2841,
+      "step": 5051
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002620547945205479,
+      "loss": 0.3205,
+      "step": 5052
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00026204234122042337,
+      "loss": 0.4672,
+      "step": 5053
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00026202988792029886,
+      "loss": 0.3333,
+      "step": 5054
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00026201743462017435,
+      "loss": 0.2539,
+      "step": 5055
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002620049813200498,
+      "loss": 0.2319,
+      "step": 5056
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00026199252801992526,
+      "loss": 0.3512,
+      "step": 5057
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002619800747198007,
+      "loss": 0.4917,
+      "step": 5058
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002619676214196762,
+      "loss": 0.4714,
+      "step": 5059
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00026195516811955167,
+      "loss": 0.2322,
+      "step": 5060
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00026194271481942715,
+      "loss": 0.2934,
+      "step": 5061
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002619302615193026,
+      "loss": 0.297,
+      "step": 5062
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00026191780821917807,
+      "loss": 0.3276,
+      "step": 5063
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002619053549190535,
+      "loss": 0.4098,
+      "step": 5064
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000261892901618929,
+      "loss": 0.3122,
+      "step": 5065
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002618804483188045,
+      "loss": 0.3475,
+      "step": 5066
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00026186799501867996,
+      "loss": 0.2749,
+      "step": 5067
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002618555417185554,
+      "loss": 0.2896,
+      "step": 5068
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00026184308841843083,
+      "loss": 0.2311,
+      "step": 5069
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002618306351183063,
+      "loss": 0.2144,
+      "step": 5070
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002618181818181818,
+      "loss": 0.3336,
+      "step": 5071
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002618057285180573,
+      "loss": 0.368,
+      "step": 5072
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002617932752179327,
+      "loss": 0.4322,
+      "step": 5073
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002617808219178082,
+      "loss": 0.2698,
+      "step": 5074
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026176836861768364,
+      "loss": 0.2652,
+      "step": 5075
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002617559153175591,
+      "loss": 0.2509,
+      "step": 5076
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002617434620174346,
+      "loss": 0.2906,
+      "step": 5077
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002617310087173101,
+      "loss": 0.6969,
+      "step": 5078
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026171855541718553,
+      "loss": 0.2532,
+      "step": 5079
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000261706102117061,
+      "loss": 0.3072,
+      "step": 5080
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026169364881693645,
+      "loss": 0.3796,
+      "step": 5081
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026168119551681193,
+      "loss": 0.208,
+      "step": 5082
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002616687422166874,
+      "loss": 0.3224,
+      "step": 5083
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026165628891656285,
+      "loss": 0.4282,
+      "step": 5084
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026164383561643834,
+      "loss": 0.271,
+      "step": 5085
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026163138231631377,
+      "loss": 0.2839,
+      "step": 5086
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026161892901618926,
+      "loss": 0.2513,
+      "step": 5087
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026160647571606474,
+      "loss": 0.3283,
+      "step": 5088
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026159402241594023,
+      "loss": 0.2384,
+      "step": 5089
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026158156911581566,
+      "loss": 0.4263,
+      "step": 5090
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026156911581569115,
+      "loss": 0.4796,
+      "step": 5091
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002615566625155666,
+      "loss": 0.2466,
+      "step": 5092
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026154420921544206,
+      "loss": 0.2951,
+      "step": 5093
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026153175591531755,
+      "loss": 0.2032,
+      "step": 5094
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026151930261519304,
+      "loss": 0.2999,
+      "step": 5095
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026150684931506847,
+      "loss": 0.3474,
+      "step": 5096
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026149439601494396,
+      "loss": 0.5935,
+      "step": 5097
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002614819427148194,
+      "loss": 0.3266,
+      "step": 5098
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002614694894146949,
+      "loss": 0.3226,
+      "step": 5099
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00026145703611457036,
+      "loss": 0.2317,
+      "step": 5100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002614445828144458,
+      "loss": 0.3075,
+      "step": 5101
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002614321295143213,
+      "loss": 0.2559,
+      "step": 5102
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002614196762141967,
+      "loss": 0.4691,
+      "step": 5103
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002614072229140722,
+      "loss": 0.2408,
+      "step": 5104
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002613947696139477,
+      "loss": 0.3266,
+      "step": 5105
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00026138231631382317,
+      "loss": 0.25,
+      "step": 5106
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002613698630136986,
+      "loss": 0.315,
+      "step": 5107
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002613574097135741,
+      "loss": 0.4884,
+      "step": 5108
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002613449564134495,
+      "loss": 0.5019,
+      "step": 5109
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.000261332503113325,
+      "loss": 0.2136,
+      "step": 5110
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002613200498132005,
+      "loss": 0.2394,
+      "step": 5111
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.000261307596513076,
+      "loss": 0.7617,
+      "step": 5112
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002612951432129514,
+      "loss": 0.2947,
+      "step": 5113
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002612826899128269,
+      "loss": 0.3733,
+      "step": 5114
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00026127023661270233,
+      "loss": 0.2415,
+      "step": 5115
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002612577833125778,
+      "loss": 0.4597,
+      "step": 5116
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002612453300124533,
+      "loss": 0.3123,
+      "step": 5117
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00026123287671232873,
+      "loss": 0.3501,
+      "step": 5118
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002612204234122042,
+      "loss": 0.3675,
+      "step": 5119
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00026120797011207965,
+      "loss": 0.3426,
+      "step": 5120
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00026119551681195514,
+      "loss": 0.2459,
+      "step": 5121
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002611830635118306,
+      "loss": 0.7319,
+      "step": 5122
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002611706102117061,
+      "loss": 0.4959,
+      "step": 5123
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00026115815691158154,
+      "loss": 0.2625,
+      "step": 5124
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00026114570361145703,
+      "loss": 0.2296,
+      "step": 5125
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00026113325031133246,
+      "loss": 0.3217,
+      "step": 5126
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00026112079701120795,
+      "loss": 0.3481,
+      "step": 5127
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00026110834371108343,
+      "loss": 0.4682,
+      "step": 5128
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002610958904109589,
+      "loss": 0.3029,
+      "step": 5129
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00026108343711083435,
+      "loss": 0.2485,
+      "step": 5130
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002610709838107098,
+      "loss": 0.2957,
+      "step": 5131
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00026105853051058527,
+      "loss": 0.3071,
+      "step": 5132
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00026104607721046076,
+      "loss": 0.3329,
+      "step": 5133
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00026103362391033624,
+      "loss": 0.486,
+      "step": 5134
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002610211706102117,
+      "loss": 0.2987,
+      "step": 5135
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00026100871731008716,
+      "loss": 0.2372,
+      "step": 5136
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002609962640099626,
+      "loss": 0.1835,
+      "step": 5137
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002609838107098381,
+      "loss": 0.3531,
+      "step": 5138
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00026097135740971357,
+      "loss": 0.3589,
+      "step": 5139
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00026095890410958905,
+      "loss": 0.2912,
+      "step": 5140
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002609464508094645,
+      "loss": 0.5084,
+      "step": 5141
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00026093399750933997,
+      "loss": 0.3735,
+      "step": 5142
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002609215442092154,
+      "loss": 0.3574,
+      "step": 5143
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002609090909090909,
+      "loss": 0.1929,
+      "step": 5144
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002608966376089664,
+      "loss": 0.3005,
+      "step": 5145
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002608841843088418,
+      "loss": 0.3394,
+      "step": 5146
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002608717310087173,
+      "loss": 0.3541,
+      "step": 5147
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002608592777085927,
+      "loss": 0.3167,
+      "step": 5148
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002608468244084682,
+      "loss": 0.2253,
+      "step": 5149
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002608343711083437,
+      "loss": 0.2197,
+      "step": 5150
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002608219178082192,
+      "loss": 0.2171,
+      "step": 5151
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002608094645080946,
+      "loss": 0.3159,
+      "step": 5152
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002607970112079701,
+      "loss": 0.3929,
+      "step": 5153
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026078455790784553,
+      "loss": 0.3529,
+      "step": 5154
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000260772104607721,
+      "loss": 0.2256,
+      "step": 5155
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002607596513075965,
+      "loss": 0.3666,
+      "step": 5156
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000260747198007472,
+      "loss": 0.1913,
+      "step": 5157
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002607347447073474,
+      "loss": 0.2432,
+      "step": 5158
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002607222914072229,
+      "loss": 0.4415,
+      "step": 5159
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026070983810709834,
+      "loss": 0.2562,
+      "step": 5160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026069738480697383,
+      "loss": 0.3908,
+      "step": 5161
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002606849315068493,
+      "loss": 0.2196,
+      "step": 5162
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026067247820672475,
+      "loss": 0.3638,
+      "step": 5163
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026066002490660023,
+      "loss": 0.3869,
+      "step": 5164
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026064757160647567,
+      "loss": 0.4149,
+      "step": 5165
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026063511830635115,
+      "loss": 0.4113,
+      "step": 5166
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026062266500622664,
+      "loss": 0.2591,
+      "step": 5167
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002606102117061021,
+      "loss": 0.2659,
+      "step": 5168
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026059775840597756,
+      "loss": 0.2436,
+      "step": 5169
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026058530510585304,
+      "loss": 0.2061,
+      "step": 5170
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002605728518057285,
+      "loss": 0.3364,
+      "step": 5171
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026056039850560396,
+      "loss": 0.5535,
+      "step": 5172
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026054794520547945,
+      "loss": 0.2907,
+      "step": 5173
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026053549190535493,
+      "loss": 0.2362,
+      "step": 5174
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026052303860523037,
+      "loss": 0.2406,
+      "step": 5175
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026051058530510585,
+      "loss": 0.2141,
+      "step": 5176
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002604981320049813,
+      "loss": 0.2475,
+      "step": 5177
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00026048567870485677,
+      "loss": 0.7304,
+      "step": 5178
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00026047322540473226,
+      "loss": 0.2258,
+      "step": 5179
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002604607721046077,
+      "loss": 0.2928,
+      "step": 5180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002604483188044832,
+      "loss": 0.2764,
+      "step": 5181
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002604358655043586,
+      "loss": 0.2923,
+      "step": 5182
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002604234122042341,
+      "loss": 0.3012,
+      "step": 5183
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002604109589041096,
+      "loss": 0.4204,
+      "step": 5184
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00026039850560398507,
+      "loss": 0.2555,
+      "step": 5185
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002603860523038605,
+      "loss": 0.293,
+      "step": 5186
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.000260373599003736,
+      "loss": 0.2217,
+      "step": 5187
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002603611457036114,
+      "loss": 0.2488,
+      "step": 5188
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002603486924034869,
+      "loss": 0.2712,
+      "step": 5189
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002603362391033624,
+      "loss": 0.34,
+      "step": 5190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002603237858032379,
+      "loss": 0.393,
+      "step": 5191
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002603113325031133,
+      "loss": 0.2195,
+      "step": 5192
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00026029887920298874,
+      "loss": 0.2852,
+      "step": 5193
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002602864259028642,
+      "loss": 0.2319,
+      "step": 5194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002602739726027397,
+      "loss": 0.2566,
+      "step": 5195
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002602615193026152,
+      "loss": 0.2743,
+      "step": 5196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00026024906600249063,
+      "loss": 0.5349,
+      "step": 5197
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002602366127023661,
+      "loss": 0.3054,
+      "step": 5198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00026022415940224155,
+      "loss": 0.2187,
+      "step": 5199
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00026021170610211704,
+      "loss": 0.2247,
+      "step": 5200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002601992528019925,
+      "loss": 0.3713,
+      "step": 5201
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.000260186799501868,
+      "loss": 0.4379,
+      "step": 5202
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00026017434620174344,
+      "loss": 0.5556,
+      "step": 5203
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002601618929016189,
+      "loss": 0.3288,
+      "step": 5204
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00026014943960149436,
+      "loss": 0.2795,
+      "step": 5205
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00026013698630136985,
+      "loss": 0.2141,
+      "step": 5206
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00026012453300124533,
+      "loss": 0.3001,
+      "step": 5207
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00026011207970112076,
+      "loss": 0.2175,
+      "step": 5208
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00026009962640099625,
+      "loss": 0.6667,
+      "step": 5209
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002600871731008717,
+      "loss": 0.2682,
+      "step": 5210
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00026007471980074717,
+      "loss": 0.3425,
+      "step": 5211
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00026006226650062265,
+      "loss": 0.2654,
+      "step": 5212
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00026004981320049814,
+      "loss": 0.2074,
+      "step": 5213
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00026003735990037357,
+      "loss": 0.2301,
+      "step": 5214
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00026002490660024906,
+      "loss": 0.3655,
+      "step": 5215
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002600124533001245,
+      "loss": 0.5022,
+      "step": 5216
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00026,
+      "loss": 0.1756,
+      "step": 5217
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00025998754669987546,
+      "loss": 0.271,
+      "step": 5218
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002599750933997509,
+      "loss": 0.3254,
+      "step": 5219
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002599626400996264,
+      "loss": 0.4154,
+      "step": 5220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00025995018679950187,
+      "loss": 0.3019,
+      "step": 5221
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002599377334993773,
+      "loss": 0.2091,
+      "step": 5222
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002599252801992528,
+      "loss": 0.3128,
+      "step": 5223
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00025991282689912827,
+      "loss": 0.2271,
+      "step": 5224
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002599003735990037,
+      "loss": 0.4489,
+      "step": 5225
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002598879202988792,
+      "loss": 0.2299,
+      "step": 5226
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002598754669987546,
+      "loss": 0.2335,
+      "step": 5227
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002598630136986301,
+      "loss": 0.2364,
+      "step": 5228
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002598505603985056,
+      "loss": 0.378,
+      "step": 5229
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00025983810709838103,
+      "loss": 0.3523,
+      "step": 5230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002598256537982565,
+      "loss": 0.432,
+      "step": 5231
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.000259813200498132,
+      "loss": 0.3057,
+      "step": 5232
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025980074719800743,
+      "loss": 0.2973,
+      "step": 5233
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002597882938978829,
+      "loss": 0.2267,
+      "step": 5234
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002597758405977584,
+      "loss": 0.2893,
+      "step": 5235
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025976338729763384,
+      "loss": 0.3029,
+      "step": 5236
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002597509339975093,
+      "loss": 0.4901,
+      "step": 5237
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002597384806973848,
+      "loss": 0.4212,
+      "step": 5238
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025972602739726024,
+      "loss": 0.2358,
+      "step": 5239
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025971357409713573,
+      "loss": 0.3388,
+      "step": 5240
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025970112079701116,
+      "loss": 0.2049,
+      "step": 5241
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025968866749688665,
+      "loss": 0.3655,
+      "step": 5242
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025967621419676213,
+      "loss": 0.4413,
+      "step": 5243
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025966376089663756,
+      "loss": 0.313,
+      "step": 5244
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025965130759651305,
+      "loss": 0.2669,
+      "step": 5245
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002596388542963885,
+      "loss": 0.2516,
+      "step": 5246
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025962640099626397,
+      "loss": 0.2042,
+      "step": 5247
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025961394769613946,
+      "loss": 0.1935,
+      "step": 5248
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025960149439601494,
+      "loss": 0.4611,
+      "step": 5249
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002595890410958904,
+      "loss": 0.449,
+      "step": 5250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025957658779576586,
+      "loss": 0.2345,
+      "step": 5251
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002595641344956413,
+      "loss": 0.2192,
+      "step": 5252
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002595516811955168,
+      "loss": 0.2188,
+      "step": 5253
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025953922789539226,
+      "loss": 0.2186,
+      "step": 5254
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00025952677459526775,
+      "loss": 0.2355,
+      "step": 5255
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002595143212951432,
+      "loss": 0.4654,
+      "step": 5256
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002595018679950186,
+      "loss": 0.2494,
+      "step": 5257
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002594894146948941,
+      "loss": 0.2663,
+      "step": 5258
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002594769613947696,
+      "loss": 0.2068,
+      "step": 5259
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002594645080946451,
+      "loss": 0.2833,
+      "step": 5260
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002594520547945205,
+      "loss": 0.1606,
+      "step": 5261
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.000259439601494396,
+      "loss": 0.5556,
+      "step": 5262
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002594271481942714,
+      "loss": 0.2586,
+      "step": 5263
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002594146948941469,
+      "loss": 0.2429,
+      "step": 5264
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002594022415940224,
+      "loss": 0.2431,
+      "step": 5265
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002593897882938979,
+      "loss": 0.1789,
+      "step": 5266
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002593773349937733,
+      "loss": 0.239,
+      "step": 5267
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002593648816936488,
+      "loss": 0.3914,
+      "step": 5268
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00025935242839352423,
+      "loss": 0.2678,
+      "step": 5269
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002593399750933997,
+      "loss": 0.2382,
+      "step": 5270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002593275217932752,
+      "loss": 0.2003,
+      "step": 5271
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00025931506849315064,
+      "loss": 0.3798,
+      "step": 5272
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002593026151930261,
+      "loss": 0.2382,
+      "step": 5273
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00025929016189290156,
+      "loss": 0.3038,
+      "step": 5274
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00025927770859277704,
+      "loss": 0.4471,
+      "step": 5275
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00025926525529265253,
+      "loss": 0.2218,
+      "step": 5276
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.000259252801992528,
+      "loss": 0.2668,
+      "step": 5277
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00025924034869240345,
+      "loss": 0.1915,
+      "step": 5278
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00025922789539227893,
+      "loss": 0.2406,
+      "step": 5279
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00025921544209215437,
+      "loss": 0.2865,
+      "step": 5280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00025920298879202985,
+      "loss": 0.4123,
+      "step": 5281
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00025919053549190534,
+      "loss": 0.2854,
+      "step": 5282
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002591780821917808,
+      "loss": 0.2741,
+      "step": 5283
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00025916562889165626,
+      "loss": 0.2628,
+      "step": 5284
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00025915317559153174,
+      "loss": 0.2819,
+      "step": 5285
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002591407222914072,
+      "loss": 0.3039,
+      "step": 5286
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00025912826899128266,
+      "loss": 0.3489,
+      "step": 5287
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00025911581569115815,
+      "loss": 0.2632,
+      "step": 5288
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002591033623910336,
+      "loss": 0.2099,
+      "step": 5289
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00025909090909090907,
+      "loss": 0.2787,
+      "step": 5290
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002590784557907845,
+      "loss": 0.1861,
+      "step": 5291
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00025906600249066,
+      "loss": 0.4267,
+      "step": 5292
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00025905354919053547,
+      "loss": 0.518,
+      "step": 5293
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00025904109589041096,
+      "loss": 0.2411,
+      "step": 5294
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002590286425902864,
+      "loss": 0.2424,
+      "step": 5295
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002590161892901619,
+      "loss": 0.3188,
+      "step": 5296
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002590037359900373,
+      "loss": 0.2887,
+      "step": 5297
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002589912826899128,
+      "loss": 0.2378,
+      "step": 5298
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002589788293897883,
+      "loss": 0.2709,
+      "step": 5299
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00025896637608966377,
+      "loss": 0.3733,
+      "step": 5300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002589539227895392,
+      "loss": 0.2654,
+      "step": 5301
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002589414694894147,
+      "loss": 0.2999,
+      "step": 5302
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002589290161892901,
+      "loss": 0.2734,
+      "step": 5303
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002589165628891656,
+      "loss": 0.3458,
+      "step": 5304
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002589041095890411,
+      "loss": 0.2727,
+      "step": 5305
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002588916562889165,
+      "loss": 0.5314,
+      "step": 5306
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.000258879202988792,
+      "loss": 0.2792,
+      "step": 5307
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00025886674968866744,
+      "loss": 0.189,
+      "step": 5308
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002588542963885429,
+      "loss": 0.2949,
+      "step": 5309
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002588418430884184,
+      "loss": 0.4881,
+      "step": 5310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002588293897882939,
+      "loss": 0.2194,
+      "step": 5311
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025881693648816933,
+      "loss": 0.3728,
+      "step": 5312
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002588044831880448,
+      "loss": 0.3926,
+      "step": 5313
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025879202988792025,
+      "loss": 0.1932,
+      "step": 5314
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025877957658779573,
+      "loss": 0.2305,
+      "step": 5315
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002587671232876712,
+      "loss": 0.3688,
+      "step": 5316
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002587546699875467,
+      "loss": 0.1903,
+      "step": 5317
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025874221668742214,
+      "loss": 0.5156,
+      "step": 5318
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025872976338729757,
+      "loss": 0.2216,
+      "step": 5319
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025871731008717306,
+      "loss": 0.2059,
+      "step": 5320
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025870485678704854,
+      "loss": 0.1861,
+      "step": 5321
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025869240348692403,
+      "loss": 0.1747,
+      "step": 5322
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025867995018679946,
+      "loss": 0.3603,
+      "step": 5323
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025866749688667495,
+      "loss": 0.3144,
+      "step": 5324
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002586550435865504,
+      "loss": 0.5123,
+      "step": 5325
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025864259028642587,
+      "loss": 0.2196,
+      "step": 5326
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025863013698630135,
+      "loss": 0.2518,
+      "step": 5327
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025861768368617684,
+      "loss": 0.3511,
+      "step": 5328
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025860523038605227,
+      "loss": 0.3991,
+      "step": 5329
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025859277708592776,
+      "loss": 0.3098,
+      "step": 5330
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002585803237858032,
+      "loss": 0.4685,
+      "step": 5331
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002585678704856787,
+      "loss": 0.4408,
+      "step": 5332
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025855541718555416,
+      "loss": 0.2564,
+      "step": 5333
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00025854296388542965,
+      "loss": 0.2327,
+      "step": 5334
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002585305105853051,
+      "loss": 0.2758,
+      "step": 5335
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002585180572851805,
+      "loss": 0.2586,
+      "step": 5336
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.000258505603985056,
+      "loss": 0.383,
+      "step": 5337
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002584931506849315,
+      "loss": 0.2596,
+      "step": 5338
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00025848069738480697,
+      "loss": 0.2926,
+      "step": 5339
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002584682440846824,
+      "loss": 0.2955,
+      "step": 5340
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002584557907845579,
+      "loss": 0.2792,
+      "step": 5341
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002584433374844333,
+      "loss": 0.2011,
+      "step": 5342
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002584308841843088,
+      "loss": 0.4453,
+      "step": 5343
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002584184308841843,
+      "loss": 0.2006,
+      "step": 5344
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002584059775840598,
+      "loss": 0.327,
+      "step": 5345
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002583935242839352,
+      "loss": 0.2446,
+      "step": 5346
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002583810709838107,
+      "loss": 0.2484,
+      "step": 5347
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00025836861768368613,
+      "loss": 0.1855,
+      "step": 5348
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002583561643835616,
+      "loss": 0.2611,
+      "step": 5349
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002583437110834371,
+      "loss": 0.2766,
+      "step": 5350
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00025833125778331254,
+      "loss": 0.2814,
+      "step": 5351
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.000258318804483188,
+      "loss": 0.2564,
+      "step": 5352
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00025830635118306345,
+      "loss": 0.2095,
+      "step": 5353
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00025829389788293894,
+      "loss": 0.1885,
+      "step": 5354
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002582814445828144,
+      "loss": 0.3767,
+      "step": 5355
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002582689912826899,
+      "loss": 0.4278,
+      "step": 5356
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00025825653798256535,
+      "loss": 0.4013,
+      "step": 5357
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00025824408468244083,
+      "loss": 0.1892,
+      "step": 5358
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00025823163138231626,
+      "loss": 0.2823,
+      "step": 5359
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00025821917808219175,
+      "loss": 0.2819,
+      "step": 5360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00025820672478206724,
+      "loss": 0.2567,
+      "step": 5361
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002581942714819427,
+      "loss": 0.5678,
+      "step": 5362
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00025818181818181815,
+      "loss": 0.3019,
+      "step": 5363
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00025816936488169364,
+      "loss": 0.176,
+      "step": 5364
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00025815691158156907,
+      "loss": 0.2304,
+      "step": 5365
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00025814445828144456,
+      "loss": 0.4313,
+      "step": 5366
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00025813200498132005,
+      "loss": 0.2846,
+      "step": 5367
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002581195516811955,
+      "loss": 0.4395,
+      "step": 5368
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00025810709838107096,
+      "loss": 0.2674,
+      "step": 5369
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002580946450809464,
+      "loss": 0.2316,
+      "step": 5370
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002580821917808219,
+      "loss": 0.2199,
+      "step": 5371
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00025806973848069737,
+      "loss": 0.2151,
+      "step": 5372
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00025805728518057285,
+      "loss": 0.2178,
+      "step": 5373
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002580448318804483,
+      "loss": 0.4069,
+      "step": 5374
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00025803237858032377,
+      "loss": 0.4423,
+      "step": 5375
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002580199252801992,
+      "loss": 0.2693,
+      "step": 5376
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002580074719800747,
+      "loss": 0.2017,
+      "step": 5377
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002579950186799502,
+      "loss": 0.2733,
+      "step": 5378
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00025798256537982566,
+      "loss": 0.2076,
+      "step": 5379
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002579701120797011,
+      "loss": 0.326,
+      "step": 5380
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002579576587795766,
+      "loss": 0.4139,
+      "step": 5381
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000257945205479452,
+      "loss": 0.2263,
+      "step": 5382
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002579327521793275,
+      "loss": 0.2455,
+      "step": 5383
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000257920298879203,
+      "loss": 0.232,
+      "step": 5384
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002579078455790784,
+      "loss": 0.332,
+      "step": 5385
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002578953922789539,
+      "loss": 0.1971,
+      "step": 5386
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00025788293897882934,
+      "loss": 0.5748,
+      "step": 5387
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0002578704856787048,
+      "loss": 0.373,
+      "step": 5388
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0002578580323785803,
+      "loss": 0.2345,
+      "step": 5389
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0002578455790784558,
+      "loss": 0.2123,
+      "step": 5390
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025783312577833123,
+      "loss": 0.2882,
+      "step": 5391
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0002578206724782067,
+      "loss": 0.2846,
+      "step": 5392
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025780821917808215,
+      "loss": 0.5179,
+      "step": 5393
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025779576587795763,
+      "loss": 0.2865,
+      "step": 5394
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0002577833125778331,
+      "loss": 0.1944,
+      "step": 5395
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0002577708592777086,
+      "loss": 0.2206,
+      "step": 5396
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025775840597758404,
+      "loss": 0.2274,
+      "step": 5397
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025774595267745947,
+      "loss": 0.2873,
+      "step": 5398
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025773349937733496,
+      "loss": 0.2421,
+      "step": 5399
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025772104607721044,
+      "loss": 0.4477,
+      "step": 5400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025770859277708593,
+      "loss": 0.3822,
+      "step": 5401
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025769613947696136,
+      "loss": 0.1919,
+      "step": 5402
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025768368617683685,
+      "loss": 0.2982,
+      "step": 5403
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0002576712328767123,
+      "loss": 0.2955,
+      "step": 5404
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025765877957658776,
+      "loss": 0.2255,
+      "step": 5405
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025764632627646325,
+      "loss": 0.4236,
+      "step": 5406
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025763387297633874,
+      "loss": 0.2154,
+      "step": 5407
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025762141967621417,
+      "loss": 0.2226,
+      "step": 5408
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025760896637608966,
+      "loss": 0.2881,
+      "step": 5409
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0002575965130759651,
+      "loss": 0.3163,
+      "step": 5410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0002575840597758406,
+      "loss": 0.2465,
+      "step": 5411
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00025757160647571606,
+      "loss": 0.3728,
+      "step": 5412
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0002575591531755915,
+      "loss": 0.2167,
+      "step": 5413
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000257546699875467,
+      "loss": 0.2131,
+      "step": 5414
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002575342465753424,
+      "loss": 0.2143,
+      "step": 5415
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002575217932752179,
+      "loss": 0.2199,
+      "step": 5416
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002575093399750934,
+      "loss": 0.2716,
+      "step": 5417
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00025749688667496887,
+      "loss": 0.4673,
+      "step": 5418
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002574844333748443,
+      "loss": 0.2147,
+      "step": 5419
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002574719800747198,
+      "loss": 0.2458,
+      "step": 5420
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002574595267745952,
+      "loss": 0.273,
+      "step": 5421
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002574470734744707,
+      "loss": 0.1593,
+      "step": 5422
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002574346201743462,
+      "loss": 0.2256,
+      "step": 5423
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002574221668742217,
+      "loss": 0.3636,
+      "step": 5424
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002574097135740971,
+      "loss": 0.3966,
+      "step": 5425
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002573972602739726,
+      "loss": 0.2872,
+      "step": 5426
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00025738480697384803,
+      "loss": 0.2181,
+      "step": 5427
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002573723536737235,
+      "loss": 0.2122,
+      "step": 5428
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000257359900373599,
+      "loss": 0.2014,
+      "step": 5429
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00025734744707347443,
+      "loss": 0.352,
+      "step": 5430
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002573349937733499,
+      "loss": 0.3786,
+      "step": 5431
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00025732254047322535,
+      "loss": 0.2055,
+      "step": 5432
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00025731008717310084,
+      "loss": 0.2011,
+      "step": 5433
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002572976338729763,
+      "loss": 0.2469,
+      "step": 5434
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0002572851805728518,
+      "loss": 0.3274,
+      "step": 5435
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00025727272727272724,
+      "loss": 0.3034,
+      "step": 5436
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00025726027397260273,
+      "loss": 0.5885,
+      "step": 5437
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00025724782067247816,
+      "loss": 0.242,
+      "step": 5438
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00025723536737235365,
+      "loss": 0.2392,
+      "step": 5439
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00025722291407222913,
+      "loss": 0.205,
+      "step": 5440
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002572104607721046,
+      "loss": 0.2921,
+      "step": 5441
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00025719800747198005,
+      "loss": 0.2232,
+      "step": 5442
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00025718555417185554,
+      "loss": 0.5018,
+      "step": 5443
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00025717310087173097,
+      "loss": 0.271,
+      "step": 5444
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00025716064757160646,
+      "loss": 0.3686,
+      "step": 5445
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00025714819427148194,
+      "loss": 0.237,
+      "step": 5446
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002571357409713574,
+      "loss": 0.1858,
+      "step": 5447
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00025712328767123286,
+      "loss": 0.1791,
+      "step": 5448
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002571108343711083,
+      "loss": 0.2062,
+      "step": 5449
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002570983810709838,
+      "loss": 0.5106,
+      "step": 5450
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00025708592777085927,
+      "loss": 0.2956,
+      "step": 5451
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00025707347447073475,
+      "loss": 0.2216,
+      "step": 5452
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002570610211706102,
+      "loss": 0.1869,
+      "step": 5453
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00025704856787048567,
+      "loss": 0.2548,
+      "step": 5454
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002570361145703611,
+      "loss": 0.3374,
+      "step": 5455
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002570236612702366,
+      "loss": 0.4803,
+      "step": 5456
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002570112079701121,
+      "loss": 0.2624,
+      "step": 5457
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00025699875466998756,
+      "loss": 0.198,
+      "step": 5458
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.000256986301369863,
+      "loss": 0.3052,
+      "step": 5459
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002569738480697384,
+      "loss": 0.4601,
+      "step": 5460
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002569613947696139,
+      "loss": 0.2313,
+      "step": 5461
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002569489414694894,
+      "loss": 0.3338,
+      "step": 5462
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002569364881693649,
+      "loss": 0.3751,
+      "step": 5463
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002569240348692403,
+      "loss": 0.1944,
+      "step": 5464
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002569115815691158,
+      "loss": 0.3279,
+      "step": 5465
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025689912826899123,
+      "loss": 0.1547,
+      "step": 5466
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002568866749688667,
+      "loss": 0.2012,
+      "step": 5467
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002568742216687422,
+      "loss": 0.3989,
+      "step": 5468
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002568617683686177,
+      "loss": 0.4031,
+      "step": 5469
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002568493150684931,
+      "loss": 0.2597,
+      "step": 5470
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002568368617683686,
+      "loss": 0.2753,
+      "step": 5471
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025682440846824404,
+      "loss": 0.3337,
+      "step": 5472
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025681195516811953,
+      "loss": 0.2406,
+      "step": 5473
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000256799501867995,
+      "loss": 0.3516,
+      "step": 5474
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002567870485678705,
+      "loss": 0.439,
+      "step": 5475
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025677459526774593,
+      "loss": 0.2253,
+      "step": 5476
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025676214196762137,
+      "loss": 0.2142,
+      "step": 5477
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025674968866749685,
+      "loss": 0.2735,
+      "step": 5478
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025673723536737234,
+      "loss": 0.2188,
+      "step": 5479
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002567247820672478,
+      "loss": 0.3316,
+      "step": 5480
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025671232876712326,
+      "loss": 0.3896,
+      "step": 5481
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025669987546699874,
+      "loss": 0.2386,
+      "step": 5482
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0002566874221668742,
+      "loss": 0.2289,
+      "step": 5483
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025667496886674966,
+      "loss": 0.3269,
+      "step": 5484
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025666251556662515,
+      "loss": 0.3146,
+      "step": 5485
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025665006226650063,
+      "loss": 0.2637,
+      "step": 5486
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025663760896637607,
+      "loss": 0.4958,
+      "step": 5487
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025662515566625155,
+      "loss": 0.2146,
+      "step": 5488
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000256612702366127,
+      "loss": 0.264,
+      "step": 5489
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025660024906600247,
+      "loss": 0.2621,
+      "step": 5490
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00025658779576587796,
+      "loss": 0.27,
+      "step": 5491
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002565753424657534,
+      "loss": 0.1803,
+      "step": 5492
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002565628891656289,
+      "loss": 0.3605,
+      "step": 5493
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002565504358655043,
+      "loss": 0.4909,
+      "step": 5494
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002565379825653798,
+      "loss": 0.2116,
+      "step": 5495
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002565255292652553,
+      "loss": 0.2675,
+      "step": 5496
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00025651307596513077,
+      "loss": 0.2793,
+      "step": 5497
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002565006226650062,
+      "loss": 0.3366,
+      "step": 5498
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002564881693648817,
+      "loss": 0.2849,
+      "step": 5499
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002564757160647571,
+      "loss": 0.4331,
+      "step": 5500
+    },
+    {
+      "epoch": 2.11,
+      "eval_cer": 0.11041613828355151,
+      "eval_loss": 0.3647739589214325,
+      "eval_runtime": 47.0659,
+      "eval_samples_per_second": 37.246,
+      "eval_steps_per_second": 4.674,
+      "eval_wer": 0.1935104312802643,
+      "step": 5500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002564632627646326,
+      "loss": 0.214,
+      "step": 5501
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002564508094645081,
+      "loss": 0.2045,
+      "step": 5502
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002564383561643836,
+      "loss": 0.2967,
+      "step": 5503
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.000256425902864259,
+      "loss": 0.2507,
+      "step": 5504
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002564134495641345,
+      "loss": 0.244,
+      "step": 5505
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002564009962640099,
+      "loss": 0.3646,
+      "step": 5506
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002563885429638854,
+      "loss": 0.2938,
+      "step": 5507
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002563760896637609,
+      "loss": 0.1827,
+      "step": 5508
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00025636363636363633,
+      "loss": 0.2864,
+      "step": 5509
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002563511830635118,
+      "loss": 0.2779,
+      "step": 5510
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00025633872976338725,
+      "loss": 0.215,
+      "step": 5511
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00025632627646326274,
+      "loss": 0.5188,
+      "step": 5512
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002563138231631382,
+      "loss": 0.2482,
+      "step": 5513
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002563013698630137,
+      "loss": 0.2422,
+      "step": 5514
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00025628891656288914,
+      "loss": 0.3099,
+      "step": 5515
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002562764632627646,
+      "loss": 0.2066,
+      "step": 5516
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00025626400996264006,
+      "loss": 0.2325,
+      "step": 5517
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00025625155666251555,
+      "loss": 0.5138,
+      "step": 5518
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00025623910336239103,
+      "loss": 0.2244,
+      "step": 5519
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002562266500622665,
+      "loss": 0.1974,
+      "step": 5520
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00025621419676214195,
+      "loss": 0.181,
+      "step": 5521
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00025620174346201744,
+      "loss": 0.305,
+      "step": 5522
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00025618929016189287,
+      "loss": 0.3156,
+      "step": 5523
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00025617683686176835,
+      "loss": 0.3037,
+      "step": 5524
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00025616438356164384,
+      "loss": 0.3115,
+      "step": 5525
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00025615193026151927,
+      "loss": 0.2023,
+      "step": 5526
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00025613947696139476,
+      "loss": 0.1823,
+      "step": 5527
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002561270236612702,
+      "loss": 0.2187,
+      "step": 5528
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002561145703611457,
+      "loss": 0.2087,
+      "step": 5529
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00025610211706102116,
+      "loss": 0.3014,
+      "step": 5530
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00025608966376089665,
+      "loss": 0.3067,
+      "step": 5531
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002560772104607721,
+      "loss": 0.2309,
+      "step": 5532
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00025606475716064757,
+      "loss": 0.2761,
+      "step": 5533
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.000256052303860523,
+      "loss": 0.1756,
+      "step": 5534
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002560398505603985,
+      "loss": 0.2428,
+      "step": 5535
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00025602739726027397,
+      "loss": 0.4084,
+      "step": 5536
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00025601494396014946,
+      "loss": 0.4126,
+      "step": 5537
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002560024906600249,
+      "loss": 0.2249,
+      "step": 5538
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002559900373599003,
+      "loss": 0.1881,
+      "step": 5539
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002559775840597758,
+      "loss": 0.2345,
+      "step": 5540
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002559651307596513,
+      "loss": 0.185,
+      "step": 5541
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002559526774595268,
+      "loss": 0.2441,
+      "step": 5542
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002559402241594022,
+      "loss": 0.5557,
+      "step": 5543
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002559277708592777,
+      "loss": 0.2174,
+      "step": 5544
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025591531755915313,
+      "loss": 0.1905,
+      "step": 5545
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002559028642590286,
+      "loss": 0.2164,
+      "step": 5546
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002558904109589041,
+      "loss": 0.2164,
+      "step": 5547
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002558779576587796,
+      "loss": 0.2091,
+      "step": 5548
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000255865504358655,
+      "loss": 0.3657,
+      "step": 5549
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002558530510585305,
+      "loss": 0.4106,
+      "step": 5550
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025584059775840594,
+      "loss": 0.3539,
+      "step": 5551
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025582814445828143,
+      "loss": 0.2745,
+      "step": 5552
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002558156911581569,
+      "loss": 0.1819,
+      "step": 5553
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025580323785803235,
+      "loss": 0.2352,
+      "step": 5554
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025579078455790783,
+      "loss": 0.3387,
+      "step": 5555
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025577833125778326,
+      "loss": 0.4089,
+      "step": 5556
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025576587795765875,
+      "loss": 0.2539,
+      "step": 5557
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025575342465753424,
+      "loss": 0.2209,
+      "step": 5558
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002557409713574097,
+      "loss": 0.1744,
+      "step": 5559
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025572851805728516,
+      "loss": 0.2079,
+      "step": 5560
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025571606475716064,
+      "loss": 0.3227,
+      "step": 5561
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002557036114570361,
+      "loss": 0.3331,
+      "step": 5562
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025569115815691156,
+      "loss": 0.1899,
+      "step": 5563
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025567870485678705,
+      "loss": 0.2192,
+      "step": 5564
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002556662515566625,
+      "loss": 0.2649,
+      "step": 5565
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025565379825653796,
+      "loss": 0.2138,
+      "step": 5566
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025564134495641345,
+      "loss": 0.3328,
+      "step": 5567
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002556288916562889,
+      "loss": 0.4469,
+      "step": 5568
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00025561643835616437,
+      "loss": 0.3846,
+      "step": 5569
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002556039850560398,
+      "loss": 0.2293,
+      "step": 5570
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002555915317559153,
+      "loss": 0.2829,
+      "step": 5571
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002555790784557908,
+      "loss": 0.2576,
+      "step": 5572
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002555666251556662,
+      "loss": 0.3286,
+      "step": 5573
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002555541718555417,
+      "loss": 0.3639,
+      "step": 5574
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002555417185554172,
+      "loss": 0.4071,
+      "step": 5575
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002555292652552926,
+      "loss": 0.2843,
+      "step": 5576
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002555168119551681,
+      "loss": 0.3756,
+      "step": 5577
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002555043586550436,
+      "loss": 0.2235,
+      "step": 5578
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.000255491905354919,
+      "loss": 0.1916,
+      "step": 5579
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002554794520547945,
+      "loss": 0.2434,
+      "step": 5580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00025546699875466993,
+      "loss": 0.3924,
+      "step": 5581
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002554545454545454,
+      "loss": 0.1903,
+      "step": 5582
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002554420921544209,
+      "loss": 0.3044,
+      "step": 5583
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002554296388542964,
+      "loss": 0.1765,
+      "step": 5584
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002554171855541718,
+      "loss": 0.2414,
+      "step": 5585
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002554047322540473,
+      "loss": 0.2129,
+      "step": 5586
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00025539227895392274,
+      "loss": 0.4813,
+      "step": 5587
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00025537982565379823,
+      "loss": 0.2582,
+      "step": 5588
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002553673723536737,
+      "loss": 0.2883,
+      "step": 5589
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00025535491905354915,
+      "loss": 0.2055,
+      "step": 5590
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00025534246575342463,
+      "loss": 0.1697,
+      "step": 5591
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00025533001245330007,
+      "loss": 0.1652,
+      "step": 5592
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00025531755915317555,
+      "loss": 0.7018,
+      "step": 5593
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00025530510585305104,
+      "loss": 0.2241,
+      "step": 5594
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0002552926525529265,
+      "loss": 0.2094,
+      "step": 5595
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00025528019925280196,
+      "loss": 0.2282,
+      "step": 5596
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00025526774595267744,
+      "loss": 0.1779,
+      "step": 5597
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002552552926525529,
+      "loss": 0.1926,
+      "step": 5598
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00025524283935242836,
+      "loss": 0.3044,
+      "step": 5599
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00025523038605230385,
+      "loss": 0.3828,
+      "step": 5600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00025521793275217933,
+      "loss": 0.2609,
+      "step": 5601
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00025520547945205477,
+      "loss": 0.254,
+      "step": 5602
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002551930261519302,
+      "loss": 0.3143,
+      "step": 5603
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002551805728518057,
+      "loss": 0.246,
+      "step": 5604
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00025516811955168117,
+      "loss": 0.2282,
+      "step": 5605
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00025515566625155666,
+      "loss": 0.5307,
+      "step": 5606
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002551432129514321,
+      "loss": 0.2204,
+      "step": 5607
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002551307596513076,
+      "loss": 0.2819,
+      "step": 5608
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.000255118306351183,
+      "loss": 0.3362,
+      "step": 5609
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002551058530510585,
+      "loss": 0.2061,
+      "step": 5610
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.000255093399750934,
+      "loss": 0.2357,
+      "step": 5611
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00025508094645080947,
+      "loss": 0.5165,
+      "step": 5612
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002550684931506849,
+      "loss": 0.2482,
+      "step": 5613
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002550560398505604,
+      "loss": 0.2261,
+      "step": 5614
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002550435865504358,
+      "loss": 0.26,
+      "step": 5615
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002550311332503113,
+      "loss": 0.3026,
+      "step": 5616
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002550186799501868,
+      "loss": 0.2018,
+      "step": 5617
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002550062266500622,
+      "loss": 0.3276,
+      "step": 5618
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002549937733499377,
+      "loss": 0.2221,
+      "step": 5619
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00025498132004981314,
+      "loss": 0.3192,
+      "step": 5620
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002549688667496886,
+      "loss": 0.2009,
+      "step": 5621
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002549564134495641,
+      "loss": 0.3346,
+      "step": 5622
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002549439601494396,
+      "loss": 0.2216,
+      "step": 5623
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025493150684931503,
+      "loss": 0.3315,
+      "step": 5624
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002549190535491905,
+      "loss": 0.4559,
+      "step": 5625
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025490660024906595,
+      "loss": 0.2065,
+      "step": 5626
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025489414694894143,
+      "loss": 0.318,
+      "step": 5627
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002548816936488169,
+      "loss": 0.2552,
+      "step": 5628
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002548692403486924,
+      "loss": 0.2093,
+      "step": 5629
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025485678704856784,
+      "loss": 0.2491,
+      "step": 5630
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002548443337484433,
+      "loss": 0.3474,
+      "step": 5631
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025483188044831876,
+      "loss": 0.255,
+      "step": 5632
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025481942714819424,
+      "loss": 0.2344,
+      "step": 5633
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025480697384806973,
+      "loss": 0.2484,
+      "step": 5634
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025479452054794516,
+      "loss": 0.2416,
+      "step": 5635
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025478206724782065,
+      "loss": 0.2949,
+      "step": 5636
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002547696139476961,
+      "loss": 0.5479,
+      "step": 5637
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025475716064757157,
+      "loss": 0.2383,
+      "step": 5638
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025474470734744705,
+      "loss": 0.2886,
+      "step": 5639
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025473225404732254,
+      "loss": 0.2549,
+      "step": 5640
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025471980074719797,
+      "loss": 0.2866,
+      "step": 5641
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025470734744707346,
+      "loss": 0.2706,
+      "step": 5642
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002546948941469489,
+      "loss": 0.5788,
+      "step": 5643
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002546824408468244,
+      "loss": 0.3439,
+      "step": 5644
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025466998754669986,
+      "loss": 0.2395,
+      "step": 5645
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025465753424657535,
+      "loss": 0.2935,
+      "step": 5646
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002546450809464508,
+      "loss": 0.1947,
+      "step": 5647
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00025463262764632627,
+      "loss": 0.3372,
+      "step": 5648
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002546201743462017,
+      "loss": 0.3094,
+      "step": 5649
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002546077210460772,
+      "loss": 0.4577,
+      "step": 5650
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00025459526774595267,
+      "loss": 0.2788,
+      "step": 5651
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002545828144458281,
+      "loss": 0.3255,
+      "step": 5652
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002545703611457036,
+      "loss": 0.2485,
+      "step": 5653
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.000254557907845579,
+      "loss": 0.1827,
+      "step": 5654
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002545454545454545,
+      "loss": 0.2609,
+      "step": 5655
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00025453300124533,
+      "loss": 0.4592,
+      "step": 5656
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002545205479452055,
+      "loss": 0.2493,
+      "step": 5657
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002545080946450809,
+      "loss": 0.3215,
+      "step": 5658
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002544956413449564,
+      "loss": 0.1989,
+      "step": 5659
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00025448318804483183,
+      "loss": 0.3584,
+      "step": 5660
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002544707347447073,
+      "loss": 0.2195,
+      "step": 5661
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002544582814445828,
+      "loss": 0.3915,
+      "step": 5662
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002544458281444583,
+      "loss": 0.3147,
+      "step": 5663
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002544333748443337,
+      "loss": 0.2351,
+      "step": 5664
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00025442092154420915,
+      "loss": 0.2289,
+      "step": 5665
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00025440846824408464,
+      "loss": 0.3369,
+      "step": 5666
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002543960149439601,
+      "loss": 0.278,
+      "step": 5667
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002543835616438356,
+      "loss": 0.3617,
+      "step": 5668
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00025437110834371105,
+      "loss": 0.2215,
+      "step": 5669
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00025435865504358653,
+      "loss": 0.3186,
+      "step": 5670
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00025434620174346196,
+      "loss": 0.2478,
+      "step": 5671
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00025433374844333745,
+      "loss": 0.1909,
+      "step": 5672
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00025432129514321294,
+      "loss": 0.2953,
+      "step": 5673
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002543088418430884,
+      "loss": 0.2883,
+      "step": 5674
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00025429638854296385,
+      "loss": 0.2962,
+      "step": 5675
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00025428393524283934,
+      "loss": 0.412,
+      "step": 5676
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00025427148194271477,
+      "loss": 0.4867,
+      "step": 5677
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00025425902864259026,
+      "loss": 0.1834,
+      "step": 5678
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00025424657534246575,
+      "loss": 0.25,
+      "step": 5679
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002542341220423412,
+      "loss": 0.5058,
+      "step": 5680
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00025422166874221666,
+      "loss": 0.5041,
+      "step": 5681
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002542092154420921,
+      "loss": 0.2518,
+      "step": 5682
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002541967621419676,
+      "loss": 0.3482,
+      "step": 5683
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00025418430884184307,
+      "loss": 0.1919,
+      "step": 5684
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00025417185554171855,
+      "loss": 0.2107,
+      "step": 5685
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.000254159402241594,
+      "loss": 0.2231,
+      "step": 5686
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00025414694894146947,
+      "loss": 0.3157,
+      "step": 5687
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002541344956413449,
+      "loss": 0.2313,
+      "step": 5688
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002541220423412204,
+      "loss": 0.1834,
+      "step": 5689
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002541095890410959,
+      "loss": 0.2147,
+      "step": 5690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00025409713574097136,
+      "loss": 0.2076,
+      "step": 5691
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002540846824408468,
+      "loss": 0.2542,
+      "step": 5692
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002540722291407223,
+      "loss": 0.4577,
+      "step": 5693
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002540597758405977,
+      "loss": 0.227,
+      "step": 5694
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002540473225404732,
+      "loss": 0.2967,
+      "step": 5695
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002540348692403487,
+      "loss": 0.3397,
+      "step": 5696
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002540224159402241,
+      "loss": 0.279,
+      "step": 5697
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002540099626400996,
+      "loss": 0.2879,
+      "step": 5698
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00025399750933997504,
+      "loss": 0.3761,
+      "step": 5699
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0002539850560398505,
+      "loss": 0.5342,
+      "step": 5700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.000253972602739726,
+      "loss": 0.2466,
+      "step": 5701
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002539601494396015,
+      "loss": 0.3604,
+      "step": 5702
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025394769613947693,
+      "loss": 0.2074,
+      "step": 5703
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002539352428393524,
+      "loss": 0.2666,
+      "step": 5704
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025392278953922785,
+      "loss": 0.2557,
+      "step": 5705
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025391033623910333,
+      "loss": 0.3218,
+      "step": 5706
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002538978829389788,
+      "loss": 0.4985,
+      "step": 5707
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002538854296388543,
+      "loss": 0.2286,
+      "step": 5708
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025387297633872974,
+      "loss": 0.2158,
+      "step": 5709
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002538605230386052,
+      "loss": 0.2196,
+      "step": 5710
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025384806973848066,
+      "loss": 0.2557,
+      "step": 5711
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025383561643835614,
+      "loss": 0.3335,
+      "step": 5712
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025382316313823163,
+      "loss": 0.3082,
+      "step": 5713
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025381070983810706,
+      "loss": 0.3426,
+      "step": 5714
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025379825653798255,
+      "loss": 0.193,
+      "step": 5715
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.000253785803237858,
+      "loss": 0.2005,
+      "step": 5716
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025377334993773346,
+      "loss": 0.2553,
+      "step": 5717
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025376089663760895,
+      "loss": 0.3891,
+      "step": 5718
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025374844333748444,
+      "loss": 0.2424,
+      "step": 5719
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025373599003735987,
+      "loss": 0.2404,
+      "step": 5720
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025372353673723536,
+      "loss": 0.2029,
+      "step": 5721
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002537110834371108,
+      "loss": 0.1969,
+      "step": 5722
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002536986301369863,
+      "loss": 0.4679,
+      "step": 5723
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025368617683686176,
+      "loss": 0.4315,
+      "step": 5724
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00025367372353673725,
+      "loss": 0.3739,
+      "step": 5725
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002536612702366127,
+      "loss": 0.216,
+      "step": 5726
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002536488169364881,
+      "loss": 0.2357,
+      "step": 5727
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002536363636363636,
+      "loss": 0.1765,
+      "step": 5728
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002536239103362391,
+      "loss": 0.2792,
+      "step": 5729
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00025361145703611457,
+      "loss": 0.2312,
+      "step": 5730
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00025359900373599,
+      "loss": 0.3644,
+      "step": 5731
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002535865504358655,
+      "loss": 0.1953,
+      "step": 5732
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002535740971357409,
+      "loss": 0.2444,
+      "step": 5733
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002535616438356164,
+      "loss": 0.2618,
+      "step": 5734
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002535491905354919,
+      "loss": 0.2345,
+      "step": 5735
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002535367372353674,
+      "loss": 0.2466,
+      "step": 5736
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002535242839352428,
+      "loss": 0.5107,
+      "step": 5737
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002535118306351183,
+      "loss": 0.2429,
+      "step": 5738
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00025349937733499373,
+      "loss": 0.365,
+      "step": 5739
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002534869240348692,
+      "loss": 0.2745,
+      "step": 5740
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002534744707347447,
+      "loss": 0.4703,
+      "step": 5741
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002534620174346202,
+      "loss": 0.3935,
+      "step": 5742
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002534495641344956,
+      "loss": 0.4921,
+      "step": 5743
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00025343711083437105,
+      "loss": 0.2643,
+      "step": 5744
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00025342465753424654,
+      "loss": 0.2392,
+      "step": 5745
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.000253412204234122,
+      "loss": 0.2916,
+      "step": 5746
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002533997509339975,
+      "loss": 0.2875,
+      "step": 5747
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00025338729763387294,
+      "loss": 0.2342,
+      "step": 5748
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00025337484433374843,
+      "loss": 0.2831,
+      "step": 5749
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00025336239103362386,
+      "loss": 0.4346,
+      "step": 5750
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00025334993773349935,
+      "loss": 0.2332,
+      "step": 5751
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00025333748443337483,
+      "loss": 0.2006,
+      "step": 5752
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002533250311332503,
+      "loss": 0.189,
+      "step": 5753
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00025331257783312575,
+      "loss": 0.2857,
+      "step": 5754
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00025330012453300124,
+      "loss": 0.3023,
+      "step": 5755
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00025328767123287667,
+      "loss": 0.4416,
+      "step": 5756
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00025327521793275216,
+      "loss": 0.3917,
+      "step": 5757
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00025326276463262764,
+      "loss": 0.1757,
+      "step": 5758
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002532503113325031,
+      "loss": 0.2491,
+      "step": 5759
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00025323785803237856,
+      "loss": 0.2643,
+      "step": 5760
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.000253225404732254,
+      "loss": 0.3592,
+      "step": 5761
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002532129514321295,
+      "loss": 0.543,
+      "step": 5762
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00025320049813200497,
+      "loss": 0.2369,
+      "step": 5763
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00025318804483188045,
+      "loss": 0.2589,
+      "step": 5764
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002531755915317559,
+      "loss": 0.1949,
+      "step": 5765
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00025316313823163137,
+      "loss": 0.1682,
+      "step": 5766
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002531506849315068,
+      "loss": 0.1914,
+      "step": 5767
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002531382316313823,
+      "loss": 0.305,
+      "step": 5768
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002531257783312578,
+      "loss": 0.2382,
+      "step": 5769
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00025311332503113326,
+      "loss": 0.2194,
+      "step": 5770
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002531008717310087,
+      "loss": 0.2024,
+      "step": 5771
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002530884184308842,
+      "loss": 0.2298,
+      "step": 5772
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002530759651307596,
+      "loss": 0.2008,
+      "step": 5773
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002530635118306351,
+      "loss": 0.4008,
+      "step": 5774
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002530510585305106,
+      "loss": 0.4165,
+      "step": 5775
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.000253038605230386,
+      "loss": 0.2851,
+      "step": 5776
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002530261519302615,
+      "loss": 0.2151,
+      "step": 5777
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00025301369863013693,
+      "loss": 0.2078,
+      "step": 5778
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002530012453300124,
+      "loss": 0.239,
+      "step": 5779
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002529887920298879,
+      "loss": 0.2525,
+      "step": 5780
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002529763387297634,
+      "loss": 0.582,
+      "step": 5781
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002529638854296388,
+      "loss": 0.2221,
+      "step": 5782
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002529514321295143,
+      "loss": 0.2412,
+      "step": 5783
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025293897882938974,
+      "loss": 0.2828,
+      "step": 5784
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025292652552926523,
+      "loss": 0.4764,
+      "step": 5785
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002529140722291407,
+      "loss": 0.3413,
+      "step": 5786
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002529016189290162,
+      "loss": 0.489,
+      "step": 5787
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025288916562889163,
+      "loss": 0.2116,
+      "step": 5788
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002528767123287671,
+      "loss": 0.3138,
+      "step": 5789
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025286425902864255,
+      "loss": 0.1807,
+      "step": 5790
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025285180572851804,
+      "loss": 0.2969,
+      "step": 5791
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002528393524283935,
+      "loss": 0.2924,
+      "step": 5792
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025282689912826896,
+      "loss": 0.2728,
+      "step": 5793
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025281444582814444,
+      "loss": 0.3785,
+      "step": 5794
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002528019925280199,
+      "loss": 0.2361,
+      "step": 5795
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025278953922789536,
+      "loss": 0.2424,
+      "step": 5796
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025277708592777085,
+      "loss": 0.1878,
+      "step": 5797
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025276463262764634,
+      "loss": 0.2147,
+      "step": 5798
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025275217932752177,
+      "loss": 0.2712,
+      "step": 5799
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025273972602739725,
+      "loss": 0.4012,
+      "step": 5800
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002527272727272727,
+      "loss": 0.2047,
+      "step": 5801
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025271481942714817,
+      "loss": 0.2068,
+      "step": 5802
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025270236612702366,
+      "loss": 0.2286,
+      "step": 5803
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00025268991282689914,
+      "loss": 0.2576,
+      "step": 5804
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002526774595267746,
+      "loss": 0.4117,
+      "step": 5805
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00025266500622665,
+      "loss": 0.345,
+      "step": 5806
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002526525529265255,
+      "loss": 0.2584,
+      "step": 5807
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.000252640099626401,
+      "loss": 0.2408,
+      "step": 5808
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00025262764632627647,
+      "loss": 0.2248,
+      "step": 5809
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002526151930261519,
+      "loss": 0.2389,
+      "step": 5810
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002526027397260274,
+      "loss": 0.2832,
+      "step": 5811
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002525902864259028,
+      "loss": 0.4048,
+      "step": 5812
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002525778331257783,
+      "loss": 0.2659,
+      "step": 5813
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002525653798256538,
+      "loss": 0.2566,
+      "step": 5814
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002525529265255293,
+      "loss": 0.1881,
+      "step": 5815
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002525404732254047,
+      "loss": 0.2223,
+      "step": 5816
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002525280199252802,
+      "loss": 0.3093,
+      "step": 5817
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002525155666251556,
+      "loss": 0.4481,
+      "step": 5818
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002525031133250311,
+      "loss": 0.2517,
+      "step": 5819
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002524906600249066,
+      "loss": 0.2456,
+      "step": 5820
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002524782067247821,
+      "loss": 0.2196,
+      "step": 5821
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002524657534246575,
+      "loss": 0.2907,
+      "step": 5822
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00025245330012453295,
+      "loss": 0.2055,
+      "step": 5823
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00025244084682440844,
+      "loss": 0.2734,
+      "step": 5824
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002524283935242839,
+      "loss": 0.4047,
+      "step": 5825
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002524159402241594,
+      "loss": 0.2139,
+      "step": 5826
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00025240348692403484,
+      "loss": 0.2796,
+      "step": 5827
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002523910336239103,
+      "loss": 0.2265,
+      "step": 5828
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00025237858032378576,
+      "loss": 0.1985,
+      "step": 5829
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00025236612702366125,
+      "loss": 0.2212,
+      "step": 5830
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00025235367372353673,
+      "loss": 0.5317,
+      "step": 5831
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002523412204234122,
+      "loss": 0.3059,
+      "step": 5832
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00025232876712328765,
+      "loss": 0.2757,
+      "step": 5833
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00025231631382316314,
+      "loss": 0.2883,
+      "step": 5834
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00025230386052303857,
+      "loss": 0.3837,
+      "step": 5835
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00025229140722291405,
+      "loss": 0.227,
+      "step": 5836
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00025227895392278954,
+      "loss": 0.2994,
+      "step": 5837
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00025226650062266497,
+      "loss": 0.2462,
+      "step": 5838
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00025225404732254046,
+      "loss": 0.2421,
+      "step": 5839
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002522415940224159,
+      "loss": 0.2005,
+      "step": 5840
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002522291407222914,
+      "loss": 0.3522,
+      "step": 5841
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00025221668742216686,
+      "loss": 0.252,
+      "step": 5842
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00025220423412204235,
+      "loss": 0.4079,
+      "step": 5843
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002521917808219178,
+      "loss": 0.2295,
+      "step": 5844
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00025217932752179327,
+      "loss": 0.3105,
+      "step": 5845
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002521668742216687,
+      "loss": 0.1886,
+      "step": 5846
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002521544209215442,
+      "loss": 0.3199,
+      "step": 5847
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002521419676214197,
+      "loss": 0.1968,
+      "step": 5848
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00025212951432129516,
+      "loss": 0.4291,
+      "step": 5849
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002521170610211706,
+      "loss": 0.2879,
+      "step": 5850
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002521046077210461,
+      "loss": 0.2516,
+      "step": 5851
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002520921544209215,
+      "loss": 0.3175,
+      "step": 5852
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.000252079701120797,
+      "loss": 0.2551,
+      "step": 5853
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002520672478206725,
+      "loss": 0.2832,
+      "step": 5854
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002520547945205479,
+      "loss": 0.1866,
+      "step": 5855
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002520423412204234,
+      "loss": 0.3065,
+      "step": 5856
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00025202988792029883,
+      "loss": 0.2282,
+      "step": 5857
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002520174346201743,
+      "loss": 0.4093,
+      "step": 5858
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002520049813200498,
+      "loss": 0.1738,
+      "step": 5859
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002519925280199253,
+      "loss": 0.1971,
+      "step": 5860
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002519800747198007,
+      "loss": 0.2356,
+      "step": 5861
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002519676214196762,
+      "loss": 0.423,
+      "step": 5862
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025195516811955164,
+      "loss": 0.2578,
+      "step": 5863
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025194271481942713,
+      "loss": 0.3112,
+      "step": 5864
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002519302615193026,
+      "loss": 0.3079,
+      "step": 5865
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002519178082191781,
+      "loss": 0.2625,
+      "step": 5866
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025190535491905353,
+      "loss": 0.2382,
+      "step": 5867
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000251892901618929,
+      "loss": 0.29,
+      "step": 5868
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025188044831880445,
+      "loss": 0.215,
+      "step": 5869
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025186799501867994,
+      "loss": 0.3235,
+      "step": 5870
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002518555417185554,
+      "loss": 0.2639,
+      "step": 5871
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025184308841843086,
+      "loss": 0.1396,
+      "step": 5872
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025183063511830634,
+      "loss": 0.2001,
+      "step": 5873
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002518181818181818,
+      "loss": 0.319,
+      "step": 5874
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025180572851805726,
+      "loss": 0.4487,
+      "step": 5875
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025179327521793275,
+      "loss": 0.386,
+      "step": 5876
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025178082191780823,
+      "loss": 0.3427,
+      "step": 5877
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025176836861768366,
+      "loss": 0.2567,
+      "step": 5878
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025175591531755915,
+      "loss": 0.2091,
+      "step": 5879
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002517434620174346,
+      "loss": 0.3516,
+      "step": 5880
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025173100871731007,
+      "loss": 0.4121,
+      "step": 5881
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00025171855541718556,
+      "loss": 0.4091,
+      "step": 5882
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000251706102117061,
+      "loss": 0.2362,
+      "step": 5883
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002516936488169365,
+      "loss": 0.3121,
+      "step": 5884
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002516811955168119,
+      "loss": 0.2899,
+      "step": 5885
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002516687422166874,
+      "loss": 0.3211,
+      "step": 5886
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002516562889165629,
+      "loss": 0.4776,
+      "step": 5887
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00025164383561643836,
+      "loss": 0.307,
+      "step": 5888
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002516313823163138,
+      "loss": 0.2124,
+      "step": 5889
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002516189290161893,
+      "loss": 0.2424,
+      "step": 5890
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002516064757160647,
+      "loss": 0.1389,
+      "step": 5891
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002515940224159402,
+      "loss": 0.1705,
+      "step": 5892
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002515815691158157,
+      "loss": 0.3566,
+      "step": 5893
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002515691158156911,
+      "loss": 0.3406,
+      "step": 5894
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002515566625155666,
+      "loss": 0.2771,
+      "step": 5895
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002515442092154421,
+      "loss": 0.2523,
+      "step": 5896
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002515317559153175,
+      "loss": 0.4473,
+      "step": 5897
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000251519302615193,
+      "loss": 0.2734,
+      "step": 5898
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002515068493150685,
+      "loss": 0.346,
+      "step": 5899
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00025149439601494393,
+      "loss": 0.3622,
+      "step": 5900
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002514819427148194,
+      "loss": 0.2476,
+      "step": 5901
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00025146948941469485,
+      "loss": 0.2445,
+      "step": 5902
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00025145703611457033,
+      "loss": 0.1998,
+      "step": 5903
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002514445828144458,
+      "loss": 0.192,
+      "step": 5904
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00025143212951432125,
+      "loss": 0.2539,
+      "step": 5905
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00025141967621419674,
+      "loss": 0.6134,
+      "step": 5906
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002514072229140722,
+      "loss": 0.2168,
+      "step": 5907
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00025139476961394766,
+      "loss": 0.2572,
+      "step": 5908
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00025138231631382314,
+      "loss": 0.2441,
+      "step": 5909
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00025136986301369863,
+      "loss": 0.2914,
+      "step": 5910
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00025135740971357406,
+      "loss": 0.2986,
+      "step": 5911
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00025134495641344955,
+      "loss": 0.2851,
+      "step": 5912
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00025133250311332503,
+      "loss": 0.1825,
+      "step": 5913
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00025132004981320047,
+      "loss": 0.4246,
+      "step": 5914
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00025130759651307595,
+      "loss": 0.2492,
+      "step": 5915
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002512951432129514,
+      "loss": 0.2732,
+      "step": 5916
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00025128268991282687,
+      "loss": 0.2578,
+      "step": 5917
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00025127023661270236,
+      "loss": 0.3991,
+      "step": 5918
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002512577833125778,
+      "loss": 0.2857,
+      "step": 5919
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002512453300124533,
+      "loss": 0.1795,
+      "step": 5920
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00025123287671232876,
+      "loss": 0.2659,
+      "step": 5921
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002512204234122042,
+      "loss": 0.3099,
+      "step": 5922
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002512079701120797,
+      "loss": 0.2464,
+      "step": 5923
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00025119551681195517,
+      "loss": 0.2107,
+      "step": 5924
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002511830635118306,
+      "loss": 0.4261,
+      "step": 5925
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002511706102117061,
+      "loss": 0.2746,
+      "step": 5926
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002511581569115815,
+      "loss": 0.3268,
+      "step": 5927
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000251145703611457,
+      "loss": 0.2134,
+      "step": 5928
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002511332503113325,
+      "loss": 0.2709,
+      "step": 5929
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000251120797011208,
+      "loss": 0.2964,
+      "step": 5930
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002511083437110834,
+      "loss": 0.3473,
+      "step": 5931
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00025109589041095884,
+      "loss": 0.2564,
+      "step": 5932
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002510834371108343,
+      "loss": 0.2575,
+      "step": 5933
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002510709838107098,
+      "loss": 0.1663,
+      "step": 5934
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002510585305105853,
+      "loss": 0.3343,
+      "step": 5935
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025104607721046073,
+      "loss": 0.2552,
+      "step": 5936
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002510336239103362,
+      "loss": 0.5717,
+      "step": 5937
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025102117061021165,
+      "loss": 0.3267,
+      "step": 5938
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025100871731008713,
+      "loss": 0.2632,
+      "step": 5939
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002509962640099626,
+      "loss": 0.2268,
+      "step": 5940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002509838107098381,
+      "loss": 0.2935,
+      "step": 5941
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025097135740971354,
+      "loss": 0.2773,
+      "step": 5942
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000250958904109589,
+      "loss": 0.3153,
+      "step": 5943
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025094645080946446,
+      "loss": 0.2801,
+      "step": 5944
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025093399750933994,
+      "loss": 0.2639,
+      "step": 5945
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025092154420921543,
+      "loss": 0.2354,
+      "step": 5946
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025090909090909086,
+      "loss": 0.1902,
+      "step": 5947
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025089663760896635,
+      "loss": 0.2732,
+      "step": 5948
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002508841843088418,
+      "loss": 0.189,
+      "step": 5949
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025087173100871727,
+      "loss": 0.3109,
+      "step": 5950
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025085927770859275,
+      "loss": 0.2916,
+      "step": 5951
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025084682440846824,
+      "loss": 0.2121,
+      "step": 5952
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025083437110834367,
+      "loss": 0.2521,
+      "step": 5953
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025082191780821916,
+      "loss": 0.4239,
+      "step": 5954
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002508094645080946,
+      "loss": 0.2168,
+      "step": 5955
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002507970112079701,
+      "loss": 0.3665,
+      "step": 5956
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025078455790784556,
+      "loss": 0.2958,
+      "step": 5957
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025077210460772105,
+      "loss": 0.3481,
+      "step": 5958
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002507596513075965,
+      "loss": 0.2337,
+      "step": 5959
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00025074719800747197,
+      "loss": 0.3608,
+      "step": 5960
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002507347447073474,
+      "loss": 0.4547,
+      "step": 5961
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002507222914072229,
+      "loss": 0.6386,
+      "step": 5962
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00025070983810709837,
+      "loss": 0.2806,
+      "step": 5963
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002506973848069738,
+      "loss": 0.2446,
+      "step": 5964
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002506849315068493,
+      "loss": 0.2869,
+      "step": 5965
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002506724782067247,
+      "loss": 0.1933,
+      "step": 5966
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002506600249066002,
+      "loss": 0.2436,
+      "step": 5967
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002506475716064757,
+      "loss": 0.4063,
+      "step": 5968
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002506351183063512,
+      "loss": 0.3151,
+      "step": 5969
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002506226650062266,
+      "loss": 0.271,
+      "step": 5970
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002506102117061021,
+      "loss": 0.2443,
+      "step": 5971
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00025059775840597753,
+      "loss": 0.441,
+      "step": 5972
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.000250585305105853,
+      "loss": 0.1774,
+      "step": 5973
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002505728518057285,
+      "loss": 0.3423,
+      "step": 5974
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.000250560398505604,
+      "loss": 0.4549,
+      "step": 5975
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002505479452054794,
+      "loss": 0.2864,
+      "step": 5976
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002505354919053549,
+      "loss": 0.2585,
+      "step": 5977
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00025052303860523034,
+      "loss": 0.2314,
+      "step": 5978
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002505105853051058,
+      "loss": 0.2741,
+      "step": 5979
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002504981320049813,
+      "loss": 0.2264,
+      "step": 5980
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00025048567870485675,
+      "loss": 0.5817,
+      "step": 5981
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00025047322540473223,
+      "loss": 0.1975,
+      "step": 5982
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00025046077210460766,
+      "loss": 0.2352,
+      "step": 5983
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00025044831880448315,
+      "loss": 0.2342,
+      "step": 5984
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00025043586550435864,
+      "loss": 0.2105,
+      "step": 5985
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002504234122042341,
+      "loss": 0.2558,
+      "step": 5986
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00025041095890410955,
+      "loss": 0.4827,
+      "step": 5987
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025039850560398504,
+      "loss": 0.3031,
+      "step": 5988
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025038605230386047,
+      "loss": 0.2124,
+      "step": 5989
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025037359900373596,
+      "loss": 0.2055,
+      "step": 5990
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025036114570361145,
+      "loss": 0.2488,
+      "step": 5991
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025034869240348693,
+      "loss": 0.1993,
+      "step": 5992
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025033623910336236,
+      "loss": 0.5241,
+      "step": 5993
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002503237858032378,
+      "loss": 0.3178,
+      "step": 5994
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002503113325031133,
+      "loss": 0.2849,
+      "step": 5995
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025029887920298877,
+      "loss": 0.2295,
+      "step": 5996
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025028642590286425,
+      "loss": 0.1522,
+      "step": 5997
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002502739726027397,
+      "loss": 0.2381,
+      "step": 5998
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002502615193026152,
+      "loss": 0.3098,
+      "step": 5999
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002502490660024906,
+      "loss": 0.3802,
+      "step": 6000
+    },
+    {
+      "epoch": 2.3,
+      "eval_cer": 0.11093509179234055,
+      "eval_loss": 0.39661991596221924,
+      "eval_runtime": 47.1413,
+      "eval_samples_per_second": 37.186,
+      "eval_steps_per_second": 4.667,
+      "eval_wer": 0.1910440331266688,
+      "step": 6000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002502366127023661,
+      "loss": 0.3134,
+      "step": 6001
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002502241594022416,
+      "loss": 0.1756,
+      "step": 6002
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025021170610211706,
+      "loss": 0.3229,
+      "step": 6003
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002501992528019925,
+      "loss": 0.1715,
+      "step": 6004
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.000250186799501868,
+      "loss": 0.3997,
+      "step": 6005
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002501743462017434,
+      "loss": 0.4687,
+      "step": 6006
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002501618929016189,
+      "loss": 0.212,
+      "step": 6007
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002501494396014944,
+      "loss": 0.2749,
+      "step": 6008
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002501369863013699,
+      "loss": 0.2366,
+      "step": 6009
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002501245330012453,
+      "loss": 0.2344,
+      "step": 6010
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025011207970112074,
+      "loss": 0.3341,
+      "step": 6011
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002500996264009962,
+      "loss": 0.3049,
+      "step": 6012
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002500871731008717,
+      "loss": 0.2075,
+      "step": 6013
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002500747198007472,
+      "loss": 0.4883,
+      "step": 6014
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00025006226650062263,
+      "loss": 0.1716,
+      "step": 6015
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002500498132004981,
+      "loss": 0.1918,
+      "step": 6016
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00025003735990037355,
+      "loss": 0.2894,
+      "step": 6017
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00025002490660024903,
+      "loss": 0.41,
+      "step": 6018
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002500124533001245,
+      "loss": 0.2832,
+      "step": 6019
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00025,
+      "loss": 0.2675,
+      "step": 6020
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024998754669987544,
+      "loss": 0.23,
+      "step": 6021
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002499750933997509,
+      "loss": 0.223,
+      "step": 6022
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024996264009962636,
+      "loss": 0.2344,
+      "step": 6023
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024995018679950184,
+      "loss": 0.3513,
+      "step": 6024
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024993773349937733,
+      "loss": 0.531,
+      "step": 6025
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024992528019925276,
+      "loss": 0.2437,
+      "step": 6026
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024991282689912825,
+      "loss": 0.2469,
+      "step": 6027
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002499003735990037,
+      "loss": 0.2087,
+      "step": 6028
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024988792029887916,
+      "loss": 0.2221,
+      "step": 6029
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024987546699875465,
+      "loss": 0.2784,
+      "step": 6030
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024986301369863014,
+      "loss": 0.3609,
+      "step": 6031
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024985056039850557,
+      "loss": 0.2271,
+      "step": 6032
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024983810709838106,
+      "loss": 0.248,
+      "step": 6033
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002498256537982565,
+      "loss": 0.2977,
+      "step": 6034
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000249813200498132,
+      "loss": 0.2231,
+      "step": 6035
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024980074719800746,
+      "loss": 0.18,
+      "step": 6036
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024978829389788295,
+      "loss": 0.4956,
+      "step": 6037
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002497758405977584,
+      "loss": 0.2468,
+      "step": 6038
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00024976338729763386,
+      "loss": 0.2459,
+      "step": 6039
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002497509339975093,
+      "loss": 0.2447,
+      "step": 6040
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002497384806973848,
+      "loss": 0.3184,
+      "step": 6041
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00024972602739726027,
+      "loss": 0.2074,
+      "step": 6042
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002497135740971357,
+      "loss": 0.5899,
+      "step": 6043
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002497011207970112,
+      "loss": 0.2308,
+      "step": 6044
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002496886674968866,
+      "loss": 0.1946,
+      "step": 6045
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002496762141967621,
+      "loss": 0.2829,
+      "step": 6046
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002496637608966376,
+      "loss": 0.2195,
+      "step": 6047
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002496513075965131,
+      "loss": 0.2283,
+      "step": 6048
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002496388542963885,
+      "loss": 0.2944,
+      "step": 6049
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.000249626400996264,
+      "loss": 0.4692,
+      "step": 6050
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00024961394769613943,
+      "loss": 0.2185,
+      "step": 6051
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002496014943960149,
+      "loss": 0.2303,
+      "step": 6052
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002495890410958904,
+      "loss": 0.1915,
+      "step": 6053
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002495765877957659,
+      "loss": 0.2194,
+      "step": 6054
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002495641344956413,
+      "loss": 0.3623,
+      "step": 6055
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002495516811955168,
+      "loss": 0.3387,
+      "step": 6056
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00024953922789539224,
+      "loss": 0.225,
+      "step": 6057
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002495267745952677,
+      "loss": 0.25,
+      "step": 6058
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002495143212951432,
+      "loss": 0.2411,
+      "step": 6059
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00024950186799501864,
+      "loss": 0.2008,
+      "step": 6060
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00024948941469489413,
+      "loss": 0.4155,
+      "step": 6061
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00024947696139476956,
+      "loss": 0.4889,
+      "step": 6062
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00024946450809464505,
+      "loss": 0.2188,
+      "step": 6063
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00024945205479452053,
+      "loss": 0.2503,
+      "step": 6064
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.000249439601494396,
+      "loss": 0.247,
+      "step": 6065
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024942714819427145,
+      "loss": 0.2232,
+      "step": 6066
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024941469489414694,
+      "loss": 0.2219,
+      "step": 6067
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024940224159402237,
+      "loss": 0.5807,
+      "step": 6068
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024938978829389786,
+      "loss": 0.2736,
+      "step": 6069
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024937733499377334,
+      "loss": 0.2222,
+      "step": 6070
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024936488169364883,
+      "loss": 0.2295,
+      "step": 6071
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024935242839352426,
+      "loss": 0.2617,
+      "step": 6072
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002493399750933997,
+      "loss": 0.2441,
+      "step": 6073
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002493275217932752,
+      "loss": 0.3537,
+      "step": 6074
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024931506849315067,
+      "loss": 0.3348,
+      "step": 6075
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024930261519302615,
+      "loss": 0.5324,
+      "step": 6076
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002492901618929016,
+      "loss": 0.3545,
+      "step": 6077
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024927770859277707,
+      "loss": 0.2239,
+      "step": 6078
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002492652552926525,
+      "loss": 0.2206,
+      "step": 6079
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.000249252801992528,
+      "loss": 0.2927,
+      "step": 6080
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002492403486924035,
+      "loss": 0.4286,
+      "step": 6081
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024922789539227896,
+      "loss": 0.2774,
+      "step": 6082
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002492154420921544,
+      "loss": 0.2035,
+      "step": 6083
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002492029887920299,
+      "loss": 0.1614,
+      "step": 6084
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002491905354919053,
+      "loss": 0.2113,
+      "step": 6085
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002491780821917808,
+      "loss": 0.3594,
+      "step": 6086
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002491656288916563,
+      "loss": 0.4169,
+      "step": 6087
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024915317559153177,
+      "loss": 0.2508,
+      "step": 6088
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002491407222914072,
+      "loss": 0.2417,
+      "step": 6089
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024912826899128263,
+      "loss": 0.3851,
+      "step": 6090
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002491158156911581,
+      "loss": 0.19,
+      "step": 6091
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002491033623910336,
+      "loss": 0.397,
+      "step": 6092
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002490909090909091,
+      "loss": 0.3743,
+      "step": 6093
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002490784557907845,
+      "loss": 0.2024,
+      "step": 6094
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024906600249066,
+      "loss": 0.2035,
+      "step": 6095
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024905354919053544,
+      "loss": 0.2705,
+      "step": 6096
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024904109589041093,
+      "loss": 0.2734,
+      "step": 6097
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002490286425902864,
+      "loss": 0.3354,
+      "step": 6098
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002490161892901619,
+      "loss": 0.2675,
+      "step": 6099
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024900373599003734,
+      "loss": 0.7833,
+      "step": 6100
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002489912826899128,
+      "loss": 0.2123,
+      "step": 6101
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024897882938978825,
+      "loss": 0.2662,
+      "step": 6102
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024896637608966374,
+      "loss": 0.3131,
+      "step": 6103
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002489539227895392,
+      "loss": 0.2079,
+      "step": 6104
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024894146948941466,
+      "loss": 0.2753,
+      "step": 6105
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024892901618929014,
+      "loss": 0.3374,
+      "step": 6106
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002489165628891656,
+      "loss": 0.204,
+      "step": 6107
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024890410958904106,
+      "loss": 0.3304,
+      "step": 6108
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024889165628891655,
+      "loss": 0.1899,
+      "step": 6109
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024887920298879204,
+      "loss": 0.193,
+      "step": 6110
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024886674968866747,
+      "loss": 0.249,
+      "step": 6111
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024885429638854295,
+      "loss": 0.5345,
+      "step": 6112
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002488418430884184,
+      "loss": 0.3359,
+      "step": 6113
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024882938978829387,
+      "loss": 0.3284,
+      "step": 6114
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024881693648816936,
+      "loss": 0.1856,
+      "step": 6115
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024880448318804484,
+      "loss": 0.1944,
+      "step": 6116
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002487920298879203,
+      "loss": 0.1805,
+      "step": 6117
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024877957658779576,
+      "loss": 0.5269,
+      "step": 6118
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002487671232876712,
+      "loss": 0.5959,
+      "step": 6119
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002487546699875467,
+      "loss": 0.2966,
+      "step": 6120
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00024874221668742217,
+      "loss": 0.1743,
+      "step": 6121
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002487297633872976,
+      "loss": 0.1913,
+      "step": 6122
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002487173100871731,
+      "loss": 0.3088,
+      "step": 6123
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002487048567870485,
+      "loss": 0.3205,
+      "step": 6124
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000248692403486924,
+      "loss": 0.4907,
+      "step": 6125
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002486799501867995,
+      "loss": 0.2261,
+      "step": 6126
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000248667496886675,
+      "loss": 0.2483,
+      "step": 6127
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002486550435865504,
+      "loss": 0.3028,
+      "step": 6128
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002486425902864259,
+      "loss": 0.3097,
+      "step": 6129
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002486301369863013,
+      "loss": 0.3413,
+      "step": 6130
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002486176836861768,
+      "loss": 0.3977,
+      "step": 6131
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002486052303860523,
+      "loss": 0.3311,
+      "step": 6132
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002485927770859278,
+      "loss": 0.1662,
+      "step": 6133
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002485803237858032,
+      "loss": 0.2516,
+      "step": 6134
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002485678704856787,
+      "loss": 0.2275,
+      "step": 6135
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00024855541718555414,
+      "loss": 0.1875,
+      "step": 6136
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002485429638854296,
+      "loss": 0.402,
+      "step": 6137
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002485305105853051,
+      "loss": 0.2304,
+      "step": 6138
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00024851805728518054,
+      "loss": 0.3465,
+      "step": 6139
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000248505603985056,
+      "loss": 0.2012,
+      "step": 6140
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00024849315068493146,
+      "loss": 0.2568,
+      "step": 6141
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00024848069738480695,
+      "loss": 0.2455,
+      "step": 6142
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00024846824408468243,
+      "loss": 0.4596,
+      "step": 6143
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002484557907845579,
+      "loss": 0.2785,
+      "step": 6144
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00024844333748443335,
+      "loss": 0.3242,
+      "step": 6145
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00024843088418430884,
+      "loss": 0.2641,
+      "step": 6146
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00024841843088418427,
+      "loss": 0.2451,
+      "step": 6147
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00024840597758405975,
+      "loss": 0.2097,
+      "step": 6148
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00024839352428393524,
+      "loss": 0.429,
+      "step": 6149
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00024838107098381073,
+      "loss": 0.4681,
+      "step": 6150
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00024836861768368616,
+      "loss": 0.2133,
+      "step": 6151
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002483561643835616,
+      "loss": 0.25,
+      "step": 6152
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002483437110834371,
+      "loss": 0.1823,
+      "step": 6153
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00024833125778331256,
+      "loss": 0.2512,
+      "step": 6154
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00024831880448318805,
+      "loss": 0.2834,
+      "step": 6155
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002483063511830635,
+      "loss": 0.5066,
+      "step": 6156
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00024829389788293897,
+      "loss": 0.2729,
+      "step": 6157
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002482814445828144,
+      "loss": 0.2233,
+      "step": 6158
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002482689912826899,
+      "loss": 0.2333,
+      "step": 6159
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002482565379825654,
+      "loss": 0.2929,
+      "step": 6160
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00024824408468244086,
+      "loss": 0.3069,
+      "step": 6161
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002482316313823163,
+      "loss": 0.357,
+      "step": 6162
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002482191780821918,
+      "loss": 0.2285,
+      "step": 6163
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002482067247820672,
+      "loss": 0.3584,
+      "step": 6164
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002481942714819427,
+      "loss": 0.2417,
+      "step": 6165
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002481818181818182,
+      "loss": 0.1707,
+      "step": 6166
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002481693648816936,
+      "loss": 0.3938,
+      "step": 6167
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002481569115815691,
+      "loss": 0.3593,
+      "step": 6168
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00024814445828144453,
+      "loss": 0.3295,
+      "step": 6169
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00024813200498132,
+      "loss": 0.3818,
+      "step": 6170
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002481195516811955,
+      "loss": 0.3724,
+      "step": 6171
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.000248107098381071,
+      "loss": 0.3111,
+      "step": 6172
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002480946450809464,
+      "loss": 0.2639,
+      "step": 6173
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002480821917808219,
+      "loss": 0.3552,
+      "step": 6174
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024806973848069734,
+      "loss": 0.411,
+      "step": 6175
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024805728518057283,
+      "loss": 0.1973,
+      "step": 6176
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002480448318804483,
+      "loss": 0.3955,
+      "step": 6177
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002480323785803238,
+      "loss": 0.2571,
+      "step": 6178
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024801992528019923,
+      "loss": 0.2565,
+      "step": 6179
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002480074719800747,
+      "loss": 0.3572,
+      "step": 6180
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024799501867995015,
+      "loss": 0.4938,
+      "step": 6181
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024798256537982564,
+      "loss": 0.2477,
+      "step": 6182
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002479701120797011,
+      "loss": 0.1996,
+      "step": 6183
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024795765877957656,
+      "loss": 0.3845,
+      "step": 6184
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024794520547945204,
+      "loss": 0.2163,
+      "step": 6185
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002479327521793275,
+      "loss": 0.3227,
+      "step": 6186
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024792029887920296,
+      "loss": 0.5386,
+      "step": 6187
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024790784557907845,
+      "loss": 0.3162,
+      "step": 6188
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024789539227895393,
+      "loss": 0.2253,
+      "step": 6189
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024788293897882936,
+      "loss": 0.2653,
+      "step": 6190
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024787048567870485,
+      "loss": 0.2198,
+      "step": 6191
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002478580323785803,
+      "loss": 0.2934,
+      "step": 6192
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024784557907845577,
+      "loss": 0.5498,
+      "step": 6193
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024783312577833126,
+      "loss": 0.2401,
+      "step": 6194
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00024782067247820674,
+      "loss": 0.2038,
+      "step": 6195
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002478082191780822,
+      "loss": 0.1671,
+      "step": 6196
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00024779576587795766,
+      "loss": 0.2233,
+      "step": 6197
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002477833125778331,
+      "loss": 0.1629,
+      "step": 6198
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002477708592777086,
+      "loss": 0.2118,
+      "step": 6199
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00024775840597758407,
+      "loss": 0.3202,
+      "step": 6200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002477459526774595,
+      "loss": 0.2066,
+      "step": 6201
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.000247733499377335,
+      "loss": 0.2462,
+      "step": 6202
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002477210460772104,
+      "loss": 0.1955,
+      "step": 6203
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002477085927770859,
+      "loss": 0.268,
+      "step": 6204
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002476961394769614,
+      "loss": 0.2312,
+      "step": 6205
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002476836861768369,
+      "loss": 0.4326,
+      "step": 6206
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002476712328767123,
+      "loss": 0.3527,
+      "step": 6207
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002476587795765878,
+      "loss": 0.2205,
+      "step": 6208
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002476463262764632,
+      "loss": 0.246,
+      "step": 6209
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002476338729763387,
+      "loss": 0.2467,
+      "step": 6210
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002476214196762142,
+      "loss": 0.3171,
+      "step": 6211
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002476089663760897,
+      "loss": 0.565,
+      "step": 6212
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002475965130759651,
+      "loss": 0.3282,
+      "step": 6213
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00024758405977584055,
+      "loss": 0.3216,
+      "step": 6214
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00024757160647571603,
+      "loss": 0.3821,
+      "step": 6215
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002475591531755915,
+      "loss": 0.2307,
+      "step": 6216
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.000247546699875467,
+      "loss": 0.2622,
+      "step": 6217
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00024753424657534244,
+      "loss": 0.406,
+      "step": 6218
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002475217932752179,
+      "loss": 0.2785,
+      "step": 6219
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00024750933997509336,
+      "loss": 0.2224,
+      "step": 6220
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00024749688667496884,
+      "loss": 0.3984,
+      "step": 6221
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00024748443337484433,
+      "loss": 0.1731,
+      "step": 6222
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002474719800747198,
+      "loss": 0.1983,
+      "step": 6223
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00024745952677459525,
+      "loss": 0.2938,
+      "step": 6224
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00024744707347447073,
+      "loss": 0.7374,
+      "step": 6225
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00024743462017434617,
+      "loss": 0.1859,
+      "step": 6226
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00024742216687422165,
+      "loss": 0.2226,
+      "step": 6227
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00024740971357409714,
+      "loss": 0.3046,
+      "step": 6228
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00024739726027397257,
+      "loss": 0.2291,
+      "step": 6229
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00024738480697384806,
+      "loss": 0.3513,
+      "step": 6230
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002473723536737235,
+      "loss": 0.5566,
+      "step": 6231
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.000247359900373599,
+      "loss": 0.273,
+      "step": 6232
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00024734744707347446,
+      "loss": 0.301,
+      "step": 6233
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00024733499377334995,
+      "loss": 0.2192,
+      "step": 6234
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002473225404732254,
+      "loss": 0.1967,
+      "step": 6235
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00024731008717310087,
+      "loss": 0.2934,
+      "step": 6236
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002472976338729763,
+      "loss": 0.3308,
+      "step": 6237
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002472851805728518,
+      "loss": 0.3054,
+      "step": 6238
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00024727272727272727,
+      "loss": 0.2204,
+      "step": 6239
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002472602739726027,
+      "loss": 0.3477,
+      "step": 6240
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002472478206724782,
+      "loss": 0.3923,
+      "step": 6241
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002472353673723537,
+      "loss": 0.269,
+      "step": 6242
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002472229140722291,
+      "loss": 0.3529,
+      "step": 6243
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002472104607721046,
+      "loss": 0.2336,
+      "step": 6244
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002471980074719801,
+      "loss": 0.3161,
+      "step": 6245
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002471855541718555,
+      "loss": 0.2255,
+      "step": 6246
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.000247173100871731,
+      "loss": 0.3578,
+      "step": 6247
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00024716064757160643,
+      "loss": 0.298,
+      "step": 6248
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002471481942714819,
+      "loss": 0.2553,
+      "step": 6249
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002471357409713574,
+      "loss": 0.2652,
+      "step": 6250
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024712328767123284,
+      "loss": 0.2502,
+      "step": 6251
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002471108343711083,
+      "loss": 0.2319,
+      "step": 6252
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002470983810709838,
+      "loss": 0.2189,
+      "step": 6253
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024708592777085924,
+      "loss": 0.3877,
+      "step": 6254
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002470734744707347,
+      "loss": 0.3117,
+      "step": 6255
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024706102117061016,
+      "loss": 0.4171,
+      "step": 6256
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024704856787048564,
+      "loss": 0.2404,
+      "step": 6257
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024703611457036113,
+      "loss": 0.2135,
+      "step": 6258
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002470236612702366,
+      "loss": 0.1713,
+      "step": 6259
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024701120797011205,
+      "loss": 0.2392,
+      "step": 6260
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024699875466998754,
+      "loss": 0.3486,
+      "step": 6261
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024698630136986297,
+      "loss": 0.4265,
+      "step": 6262
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024697384806973845,
+      "loss": 0.3054,
+      "step": 6263
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024696139476961394,
+      "loss": 0.2604,
+      "step": 6264
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024694894146948937,
+      "loss": 0.3168,
+      "step": 6265
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024693648816936486,
+      "loss": 0.2778,
+      "step": 6266
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002469240348692403,
+      "loss": 0.2375,
+      "step": 6267
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002469115815691158,
+      "loss": 0.5087,
+      "step": 6268
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024689912826899126,
+      "loss": 0.2301,
+      "step": 6269
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024688667496886675,
+      "loss": 0.2095,
+      "step": 6270
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002468742216687422,
+      "loss": 0.187,
+      "step": 6271
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024686176836861767,
+      "loss": 0.2616,
+      "step": 6272
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002468493150684931,
+      "loss": 0.2408,
+      "step": 6273
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002468368617683686,
+      "loss": 0.317,
+      "step": 6274
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00024682440846824407,
+      "loss": 0.3877,
+      "step": 6275
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00024681195516811956,
+      "loss": 0.363,
+      "step": 6276
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.000246799501867995,
+      "loss": 0.2035,
+      "step": 6277
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002467870485678704,
+      "loss": 0.2491,
+      "step": 6278
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002467745952677459,
+      "loss": 0.2262,
+      "step": 6279
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002467621419676214,
+      "loss": 0.2065,
+      "step": 6280
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002467496886674969,
+      "loss": 0.414,
+      "step": 6281
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002467372353673723,
+      "loss": 0.3093,
+      "step": 6282
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002467247820672478,
+      "loss": 0.3886,
+      "step": 6283
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00024671232876712323,
+      "loss": 0.326,
+      "step": 6284
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002466998754669987,
+      "loss": 0.1989,
+      "step": 6285
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002466874221668742,
+      "loss": 0.2151,
+      "step": 6286
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002466749688667497,
+      "loss": 0.4838,
+      "step": 6287
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002466625155666251,
+      "loss": 0.3095,
+      "step": 6288
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002466500622665006,
+      "loss": 0.204,
+      "step": 6289
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00024663760896637604,
+      "loss": 0.2452,
+      "step": 6290
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002466251556662515,
+      "loss": 0.1821,
+      "step": 6291
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.000246612702366127,
+      "loss": 0.1982,
+      "step": 6292
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00024660024906600245,
+      "loss": 0.3768,
+      "step": 6293
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00024658779576587793,
+      "loss": 0.6305,
+      "step": 6294
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00024657534246575336,
+      "loss": 0.3133,
+      "step": 6295
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00024656288916562885,
+      "loss": 0.2634,
+      "step": 6296
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00024655043586550434,
+      "loss": 0.3039,
+      "step": 6297
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002465379825653798,
+      "loss": 0.2214,
+      "step": 6298
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00024652552926525525,
+      "loss": 0.206,
+      "step": 6299
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00024651307596513074,
+      "loss": 0.455,
+      "step": 6300
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002465006226650062,
+      "loss": 0.2838,
+      "step": 6301
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00024648816936488166,
+      "loss": 0.2736,
+      "step": 6302
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00024647571606475715,
+      "loss": 0.2327,
+      "step": 6303
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00024646326276463263,
+      "loss": 0.2346,
+      "step": 6304
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00024645080946450806,
+      "loss": 0.3375,
+      "step": 6305
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00024643835616438355,
+      "loss": 0.5718,
+      "step": 6306
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.000246425902864259,
+      "loss": 0.4479,
+      "step": 6307
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00024641344956413447,
+      "loss": 0.192,
+      "step": 6308
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00024640099626400995,
+      "loss": 0.244,
+      "step": 6309
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002463885429638854,
+      "loss": 0.2706,
+      "step": 6310
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002463760896637609,
+      "loss": 0.1803,
+      "step": 6311
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002463636363636363,
+      "loss": 0.4129,
+      "step": 6312
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002463511830635118,
+      "loss": 0.2553,
+      "step": 6313
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002463387297633873,
+      "loss": 0.2227,
+      "step": 6314
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00024632627646326276,
+      "loss": 0.1821,
+      "step": 6315
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002463138231631382,
+      "loss": 0.1993,
+      "step": 6316
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002463013698630137,
+      "loss": 0.3349,
+      "step": 6317
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002462889165628891,
+      "loss": 0.4287,
+      "step": 6318
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002462764632627646,
+      "loss": 0.3319,
+      "step": 6319
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002462640099626401,
+      "loss": 0.3247,
+      "step": 6320
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002462515566625156,
+      "loss": 0.2278,
+      "step": 6321
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.000246239103362391,
+      "loss": 0.2855,
+      "step": 6322
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002462266500622665,
+      "loss": 0.3963,
+      "step": 6323
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002462141967621419,
+      "loss": 0.2123,
+      "step": 6324
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002462017434620174,
+      "loss": 0.4314,
+      "step": 6325
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002461892901618929,
+      "loss": 0.2935,
+      "step": 6326
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024617683686176833,
+      "loss": 0.2414,
+      "step": 6327
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002461643835616438,
+      "loss": 0.2671,
+      "step": 6328
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024615193026151925,
+      "loss": 0.1927,
+      "step": 6329
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024613947696139473,
+      "loss": 0.4394,
+      "step": 6330
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002461270236612702,
+      "loss": 0.6022,
+      "step": 6331
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002461145703611457,
+      "loss": 0.2614,
+      "step": 6332
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024610211706102114,
+      "loss": 0.3754,
+      "step": 6333
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002460896637608966,
+      "loss": 0.2097,
+      "step": 6334
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024607721046077206,
+      "loss": 0.2149,
+      "step": 6335
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024606475716064754,
+      "loss": 0.2314,
+      "step": 6336
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024605230386052303,
+      "loss": 0.3145,
+      "step": 6337
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002460398505603985,
+      "loss": 0.2318,
+      "step": 6338
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024602739726027395,
+      "loss": 0.28,
+      "step": 6339
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002460149439601494,
+      "loss": 0.2255,
+      "step": 6340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024600249066002486,
+      "loss": 0.1625,
+      "step": 6341
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024599003735990035,
+      "loss": 0.2015,
+      "step": 6342
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024597758405977584,
+      "loss": 0.2582,
+      "step": 6343
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024596513075965127,
+      "loss": 0.2828,
+      "step": 6344
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024595267745952676,
+      "loss": 0.1863,
+      "step": 6345
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002459402241594022,
+      "loss": 0.243,
+      "step": 6346
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002459277708592777,
+      "loss": 0.2689,
+      "step": 6347
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024591531755915316,
+      "loss": 0.2816,
+      "step": 6348
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024590286425902865,
+      "loss": 0.3053,
+      "step": 6349
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002458904109589041,
+      "loss": 0.3334,
+      "step": 6350
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024587795765877957,
+      "loss": 0.2487,
+      "step": 6351
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.000245865504358655,
+      "loss": 0.317,
+      "step": 6352
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002458530510585305,
+      "loss": 0.3497,
+      "step": 6353
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00024584059775840597,
+      "loss": 0.2153,
+      "step": 6354
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00024582814445828146,
+      "loss": 0.224,
+      "step": 6355
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002458156911581569,
+      "loss": 0.3915,
+      "step": 6356
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002458032378580323,
+      "loss": 0.2077,
+      "step": 6357
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002457907845579078,
+      "loss": 0.2114,
+      "step": 6358
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002457783312577833,
+      "loss": 0.2993,
+      "step": 6359
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002457658779576588,
+      "loss": 0.3031,
+      "step": 6360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002457534246575342,
+      "loss": 0.2381,
+      "step": 6361
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002457409713574097,
+      "loss": 0.2606,
+      "step": 6362
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00024572851805728513,
+      "loss": 0.3796,
+      "step": 6363
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002457160647571606,
+      "loss": 0.2317,
+      "step": 6364
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002457036114570361,
+      "loss": 0.2055,
+      "step": 6365
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002456911581569116,
+      "loss": 0.2079,
+      "step": 6366
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.000245678704856787,
+      "loss": 0.3028,
+      "step": 6367
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002456662515566625,
+      "loss": 0.2594,
+      "step": 6368
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00024565379825653794,
+      "loss": 0.2236,
+      "step": 6369
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002456413449564134,
+      "loss": 0.2134,
+      "step": 6370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002456288916562889,
+      "loss": 0.187,
+      "step": 6371
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00024561643835616434,
+      "loss": 0.2054,
+      "step": 6372
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00024560398505603983,
+      "loss": 0.2612,
+      "step": 6373
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00024559153175591526,
+      "loss": 0.3998,
+      "step": 6374
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00024557907845579075,
+      "loss": 0.3783,
+      "step": 6375
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00024556662515566623,
+      "loss": 0.291,
+      "step": 6376
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002455541718555417,
+      "loss": 0.2487,
+      "step": 6377
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00024554171855541715,
+      "loss": 0.1976,
+      "step": 6378
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00024552926525529264,
+      "loss": 0.2308,
+      "step": 6379
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00024551681195516807,
+      "loss": 0.2595,
+      "step": 6380
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00024550435865504356,
+      "loss": 0.5282,
+      "step": 6381
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00024549190535491904,
+      "loss": 0.2631,
+      "step": 6382
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00024547945205479453,
+      "loss": 0.3627,
+      "step": 6383
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00024546699875466996,
+      "loss": 0.2362,
+      "step": 6384
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00024545454545454545,
+      "loss": 0.2542,
+      "step": 6385
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002454420921544209,
+      "loss": 0.3903,
+      "step": 6386
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00024542963885429637,
+      "loss": 0.3411,
+      "step": 6387
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00024541718555417185,
+      "loss": 0.2636,
+      "step": 6388
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002454047322540473,
+      "loss": 0.3954,
+      "step": 6389
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00024539227895392277,
+      "loss": 0.2745,
+      "step": 6390
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002453798256537982,
+      "loss": 0.1923,
+      "step": 6391
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002453673723536737,
+      "loss": 0.1738,
+      "step": 6392
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002453549190535492,
+      "loss": 0.4951,
+      "step": 6393
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00024534246575342466,
+      "loss": 0.264,
+      "step": 6394
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002453300124533001,
+      "loss": 0.3712,
+      "step": 6395
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002453175591531756,
+      "loss": 0.2407,
+      "step": 6396
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.000245305105853051,
+      "loss": 0.2331,
+      "step": 6397
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002452926525529265,
+      "loss": 0.3076,
+      "step": 6398
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.000245280199252802,
+      "loss": 0.3524,
+      "step": 6399
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00024526774595267747,
+      "loss": 0.3247,
+      "step": 6400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002452552926525529,
+      "loss": 0.2673,
+      "step": 6401
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002452428393524284,
+      "loss": 0.2166,
+      "step": 6402
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002452303860523038,
+      "loss": 0.2327,
+      "step": 6403
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002452179327521793,
+      "loss": 0.2176,
+      "step": 6404
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002452054794520548,
+      "loss": 0.38,
+      "step": 6405
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002451930261519302,
+      "loss": 0.2841,
+      "step": 6406
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002451805728518057,
+      "loss": 0.2882,
+      "step": 6407
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024516811955168114,
+      "loss": 0.1985,
+      "step": 6408
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024515566625155663,
+      "loss": 0.191,
+      "step": 6409
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002451432129514321,
+      "loss": 0.2788,
+      "step": 6410
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002451307596513076,
+      "loss": 0.218,
+      "step": 6411
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002451307596513076,
+      "loss": 3.616,
+      "step": 6412
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024511830635118304,
+      "loss": 0.2499,
+      "step": 6413
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002451058530510585,
+      "loss": 0.296,
+      "step": 6414
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024509339975093395,
+      "loss": 0.3962,
+      "step": 6415
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024508094645080944,
+      "loss": 0.2971,
+      "step": 6416
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002450684931506849,
+      "loss": 0.3409,
+      "step": 6417
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002450560398505604,
+      "loss": 0.402,
+      "step": 6418
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024504358655043584,
+      "loss": 0.3921,
+      "step": 6419
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002450311332503113,
+      "loss": 0.2436,
+      "step": 6420
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024501867995018676,
+      "loss": 0.2822,
+      "step": 6421
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024500622665006225,
+      "loss": 0.257,
+      "step": 6422
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024499377334993774,
+      "loss": 0.3138,
+      "step": 6423
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024498132004981317,
+      "loss": 0.3783,
+      "step": 6424
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024496886674968865,
+      "loss": 0.3658,
+      "step": 6425
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002449564134495641,
+      "loss": 0.2737,
+      "step": 6426
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024494396014943957,
+      "loss": 0.1728,
+      "step": 6427
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024493150684931506,
+      "loss": 0.3532,
+      "step": 6428
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024491905354919054,
+      "loss": 0.3724,
+      "step": 6429
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000244906600249066,
+      "loss": 0.2077,
+      "step": 6430
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024489414694894146,
+      "loss": 0.5662,
+      "step": 6431
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002448816936488169,
+      "loss": 0.2871,
+      "step": 6432
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002448692403486924,
+      "loss": 0.2517,
+      "step": 6433
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00024485678704856787,
+      "loss": 0.2833,
+      "step": 6434
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002448443337484433,
+      "loss": 0.2203,
+      "step": 6435
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002448318804483188,
+      "loss": 0.2719,
+      "step": 6436
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002448194271481942,
+      "loss": 0.4869,
+      "step": 6437
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002448069738480697,
+      "loss": 0.2816,
+      "step": 6438
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002447945205479452,
+      "loss": 0.2362,
+      "step": 6439
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002447820672478207,
+      "loss": 0.2093,
+      "step": 6440
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002447696139476961,
+      "loss": 0.1995,
+      "step": 6441
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002447571606475716,
+      "loss": 0.2543,
+      "step": 6442
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.000244744707347447,
+      "loss": 0.422,
+      "step": 6443
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002447322540473225,
+      "loss": 0.1789,
+      "step": 6444
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.000244719800747198,
+      "loss": 0.2082,
+      "step": 6445
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002447073474470735,
+      "loss": 0.2029,
+      "step": 6446
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002446948941469489,
+      "loss": 0.2962,
+      "step": 6447
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002446824408468244,
+      "loss": 0.5539,
+      "step": 6448
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00024466998754669984,
+      "loss": 0.3224,
+      "step": 6449
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002446575342465753,
+      "loss": 0.3603,
+      "step": 6450
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002446450809464508,
+      "loss": 0.3294,
+      "step": 6451
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00024463262764632624,
+      "loss": 0.2764,
+      "step": 6452
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00024462017434620173,
+      "loss": 0.2639,
+      "step": 6453
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00024460772104607716,
+      "loss": 0.2175,
+      "step": 6454
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00024459526774595265,
+      "loss": 0.2835,
+      "step": 6455
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00024458281444582813,
+      "loss": 0.3226,
+      "step": 6456
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002445703611457036,
+      "loss": 0.2105,
+      "step": 6457
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024455790784557905,
+      "loss": 0.2124,
+      "step": 6458
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024454545454545454,
+      "loss": 0.2632,
+      "step": 6459
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024453300124532997,
+      "loss": 0.2476,
+      "step": 6460
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024452054794520545,
+      "loss": 0.3023,
+      "step": 6461
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024450809464508094,
+      "loss": 0.3623,
+      "step": 6462
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024449564134495643,
+      "loss": 0.25,
+      "step": 6463
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024448318804483186,
+      "loss": 0.2641,
+      "step": 6464
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024447073474470735,
+      "loss": 0.233,
+      "step": 6465
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002444582814445828,
+      "loss": 0.2168,
+      "step": 6466
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024444582814445826,
+      "loss": 0.2553,
+      "step": 6467
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024443337484433375,
+      "loss": 0.4773,
+      "step": 6468
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002444209215442092,
+      "loss": 0.3685,
+      "step": 6469
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024440846824408467,
+      "loss": 0.2299,
+      "step": 6470
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002443960149439601,
+      "loss": 0.1787,
+      "step": 6471
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002443835616438356,
+      "loss": 0.2216,
+      "step": 6472
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002443711083437111,
+      "loss": 0.2252,
+      "step": 6473
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024435865504358656,
+      "loss": 0.2314,
+      "step": 6474
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000244346201743462,
+      "loss": 0.3298,
+      "step": 6475
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002443337484433375,
+      "loss": 0.2211,
+      "step": 6476
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002443212951432129,
+      "loss": 0.2105,
+      "step": 6477
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002443088418430884,
+      "loss": 0.2574,
+      "step": 6478
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002442963885429639,
+      "loss": 0.1602,
+      "step": 6479
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024428393524283937,
+      "loss": 0.2571,
+      "step": 6480
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002442714819427148,
+      "loss": 0.6209,
+      "step": 6481
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00024425902864259023,
+      "loss": 0.633,
+      "step": 6482
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002442465753424657,
+      "loss": 0.2617,
+      "step": 6483
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002442341220423412,
+      "loss": 0.1743,
+      "step": 6484
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002442216687422167,
+      "loss": 0.1939,
+      "step": 6485
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002442092154420921,
+      "loss": 0.2429,
+      "step": 6486
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002441967621419676,
+      "loss": 0.3971,
+      "step": 6487
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024418430884184304,
+      "loss": 0.3028,
+      "step": 6488
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024417185554171853,
+      "loss": 0.2528,
+      "step": 6489
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.000244159402241594,
+      "loss": 0.2504,
+      "step": 6490
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002441469489414695,
+      "loss": 0.2156,
+      "step": 6491
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024413449564134493,
+      "loss": 0.2871,
+      "step": 6492
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024412204234122042,
+      "loss": 0.5182,
+      "step": 6493
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024410958904109588,
+      "loss": 0.272,
+      "step": 6494
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024409713574097134,
+      "loss": 0.2287,
+      "step": 6495
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002440846824408468,
+      "loss": 0.2717,
+      "step": 6496
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024407222914072228,
+      "loss": 0.2114,
+      "step": 6497
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024405977584059774,
+      "loss": 0.2862,
+      "step": 6498
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002440473225404732,
+      "loss": 0.282,
+      "step": 6499
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024403486924034866,
+      "loss": 0.3928,
+      "step": 6500
+    },
+    {
+      "epoch": 2.49,
+      "eval_cer": 0.11004935911192618,
+      "eval_loss": 0.39953118562698364,
+      "eval_runtime": 46.8703,
+      "eval_samples_per_second": 37.401,
+      "eval_steps_per_second": 4.694,
+      "eval_wer": 0.1897995202968729,
+      "step": 6500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024402241594022412,
+      "loss": 0.3229,
+      "step": 6501
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002440099626400996,
+      "loss": 0.3697,
+      "step": 6502
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024399750933997507,
+      "loss": 0.3533,
+      "step": 6503
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024398505603985055,
+      "loss": 0.2342,
+      "step": 6504
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.000243972602739726,
+      "loss": 0.3245,
+      "step": 6505
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024396014943960147,
+      "loss": 0.487,
+      "step": 6506
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024394769613947693,
+      "loss": 0.2351,
+      "step": 6507
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024393524283935242,
+      "loss": 0.2332,
+      "step": 6508
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024392278953922787,
+      "loss": 0.2655,
+      "step": 6509
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024391033623910336,
+      "loss": 0.2193,
+      "step": 6510
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002438978829389788,
+      "loss": 0.266,
+      "step": 6511
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024388542963885428,
+      "loss": 0.5053,
+      "step": 6512
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024387297633872974,
+      "loss": 0.2109,
+      "step": 6513
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002438605230386052,
+      "loss": 0.3845,
+      "step": 6514
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024384806973848068,
+      "loss": 0.326,
+      "step": 6515
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024383561643835614,
+      "loss": 0.1948,
+      "step": 6516
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002438231631382316,
+      "loss": 0.2307,
+      "step": 6517
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024381070983810706,
+      "loss": 0.3253,
+      "step": 6518
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024379825653798255,
+      "loss": 0.2817,
+      "step": 6519
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.000243785803237858,
+      "loss": 0.2782,
+      "step": 6520
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002437733499377335,
+      "loss": 0.2203,
+      "step": 6521
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024376089663760892,
+      "loss": 0.3233,
+      "step": 6522
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002437484433374844,
+      "loss": 0.1711,
+      "step": 6523
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024373599003735987,
+      "loss": 0.3293,
+      "step": 6524
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024372353673723536,
+      "loss": 0.4787,
+      "step": 6525
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024371108343711082,
+      "loss": 0.455,
+      "step": 6526
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002436986301369863,
+      "loss": 0.1998,
+      "step": 6527
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024368617683686173,
+      "loss": 0.3112,
+      "step": 6528
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002436737235367372,
+      "loss": 0.3312,
+      "step": 6529
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024366127023661268,
+      "loss": 0.281,
+      "step": 6530
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024364881693648814,
+      "loss": 0.3919,
+      "step": 6531
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024363636363636362,
+      "loss": 0.3062,
+      "step": 6532
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024362391033623906,
+      "loss": 0.1816,
+      "step": 6533
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024361145703611454,
+      "loss": 0.2446,
+      "step": 6534
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00024359900373599,
+      "loss": 0.2418,
+      "step": 6535
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0002435865504358655,
+      "loss": 0.2443,
+      "step": 6536
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024357409713574095,
+      "loss": 0.4381,
+      "step": 6537
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024356164383561643,
+      "loss": 0.2048,
+      "step": 6538
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024354919053549187,
+      "loss": 0.3121,
+      "step": 6539
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024353673723536735,
+      "loss": 0.3716,
+      "step": 6540
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0002435242839352428,
+      "loss": 0.19,
+      "step": 6541
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0002435118306351183,
+      "loss": 0.2153,
+      "step": 6542
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024349937733499376,
+      "loss": 0.4405,
+      "step": 6543
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024348692403486924,
+      "loss": 0.3605,
+      "step": 6544
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024347447073474468,
+      "loss": 0.3528,
+      "step": 6545
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024346201743462013,
+      "loss": 0.1971,
+      "step": 6546
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024344956413449562,
+      "loss": 0.3,
+      "step": 6547
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024343711083437108,
+      "loss": 0.3,
+      "step": 6548
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024342465753424657,
+      "loss": 0.4353,
+      "step": 6549
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.000243412204234122,
+      "loss": 0.3783,
+      "step": 6550
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024339975093399748,
+      "loss": 0.2507,
+      "step": 6551
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024338729763387294,
+      "loss": 0.2221,
+      "step": 6552
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024337484433374843,
+      "loss": 0.2993,
+      "step": 6553
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0002433623910336239,
+      "loss": 0.2377,
+      "step": 6554
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024334993773349938,
+      "loss": 0.2379,
+      "step": 6555
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0002433374844333748,
+      "loss": 0.4455,
+      "step": 6556
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0002433250311332503,
+      "loss": 0.2207,
+      "step": 6557
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024331257783312575,
+      "loss": 0.3118,
+      "step": 6558
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024330012453300124,
+      "loss": 0.2297,
+      "step": 6559
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0002432876712328767,
+      "loss": 0.3659,
+      "step": 6560
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00024327521793275213,
+      "loss": 0.1997,
+      "step": 6561
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024326276463262762,
+      "loss": 0.4032,
+      "step": 6562
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024325031133250308,
+      "loss": 0.3358,
+      "step": 6563
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024323785803237856,
+      "loss": 0.5851,
+      "step": 6564
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024322540473225402,
+      "loss": 0.2542,
+      "step": 6565
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0002432129514321295,
+      "loss": 0.2005,
+      "step": 6566
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024320049813200494,
+      "loss": 0.2202,
+      "step": 6567
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024318804483188043,
+      "loss": 0.3892,
+      "step": 6568
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024317559153175589,
+      "loss": 0.2818,
+      "step": 6569
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024316313823163137,
+      "loss": 0.229,
+      "step": 6570
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024315068493150683,
+      "loss": 0.2909,
+      "step": 6571
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024313823163138232,
+      "loss": 0.1916,
+      "step": 6572
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024312577833125775,
+      "loss": 0.2296,
+      "step": 6573
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024311332503113324,
+      "loss": 0.2585,
+      "step": 6574
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0002431008717310087,
+      "loss": 0.3573,
+      "step": 6575
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024308841843088415,
+      "loss": 0.2423,
+      "step": 6576
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024307596513075964,
+      "loss": 0.317,
+      "step": 6577
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024306351183063507,
+      "loss": 0.2139,
+      "step": 6578
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024305105853051056,
+      "loss": 0.2741,
+      "step": 6579
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024303860523038602,
+      "loss": 0.2173,
+      "step": 6580
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0002430261519302615,
+      "loss": 0.4681,
+      "step": 6581
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024301369863013696,
+      "loss": 0.2771,
+      "step": 6582
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024300124533001245,
+      "loss": 0.1818,
+      "step": 6583
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024298879202988788,
+      "loss": 0.273,
+      "step": 6584
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024297633872976337,
+      "loss": 0.2338,
+      "step": 6585
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00024296388542963883,
+      "loss": 0.2262,
+      "step": 6586
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0002429514321295143,
+      "loss": 0.5285,
+      "step": 6587
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024293897882938977,
+      "loss": 0.3377,
+      "step": 6588
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024292652552926526,
+      "loss": 0.2462,
+      "step": 6589
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002429140722291407,
+      "loss": 0.2188,
+      "step": 6590
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024290161892901618,
+      "loss": 0.2207,
+      "step": 6591
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024288916562889164,
+      "loss": 0.2369,
+      "step": 6592
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002428767123287671,
+      "loss": 0.3366,
+      "step": 6593
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024286425902864258,
+      "loss": 0.322,
+      "step": 6594
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.000242851805728518,
+      "loss": 0.2033,
+      "step": 6595
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002428393524283935,
+      "loss": 0.2322,
+      "step": 6596
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024282689912826896,
+      "loss": 0.1652,
+      "step": 6597
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024281444582814445,
+      "loss": 0.242,
+      "step": 6598
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002428019925280199,
+      "loss": 0.4149,
+      "step": 6599
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002427895392278954,
+      "loss": 0.5725,
+      "step": 6600
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024277708592777082,
+      "loss": 0.2169,
+      "step": 6601
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002427646326276463,
+      "loss": 0.2264,
+      "step": 6602
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024275217932752177,
+      "loss": 0.257,
+      "step": 6603
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024273972602739725,
+      "loss": 0.3794,
+      "step": 6604
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002427272727272727,
+      "loss": 0.2344,
+      "step": 6605
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002427148194271482,
+      "loss": 0.4372,
+      "step": 6606
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024270236612702363,
+      "loss": 0.2412,
+      "step": 6607
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002426899128268991,
+      "loss": 0.2887,
+      "step": 6608
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024267745952677458,
+      "loss": 0.1786,
+      "step": 6609
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024266500622665004,
+      "loss": 0.2377,
+      "step": 6610
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024265255292652552,
+      "loss": 0.2746,
+      "step": 6611
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024264009962640095,
+      "loss": 0.3784,
+      "step": 6612
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00024262764632627644,
+      "loss": 0.2996,
+      "step": 6613
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002426151930261519,
+      "loss": 0.228,
+      "step": 6614
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024260273972602739,
+      "loss": 0.3057,
+      "step": 6615
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024259028642590285,
+      "loss": 0.1577,
+      "step": 6616
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024257783312577833,
+      "loss": 0.3061,
+      "step": 6617
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024256537982565376,
+      "loss": 0.3299,
+      "step": 6618
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024255292652552925,
+      "loss": 0.2595,
+      "step": 6619
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002425404732254047,
+      "loss": 0.3693,
+      "step": 6620
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002425280199252802,
+      "loss": 0.1917,
+      "step": 6621
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024251556662515565,
+      "loss": 0.1998,
+      "step": 6622
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002425031133250311,
+      "loss": 0.1968,
+      "step": 6623
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024249066002490657,
+      "loss": 0.4709,
+      "step": 6624
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024247820672478203,
+      "loss": 0.3406,
+      "step": 6625
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024246575342465752,
+      "loss": 0.3768,
+      "step": 6626
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024245330012453298,
+      "loss": 0.3095,
+      "step": 6627
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024244084682440846,
+      "loss": 0.1996,
+      "step": 6628
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002424283935242839,
+      "loss": 0.3274,
+      "step": 6629
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024241594022415938,
+      "loss": 0.2196,
+      "step": 6630
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024240348692403484,
+      "loss": 0.5567,
+      "step": 6631
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024239103362391033,
+      "loss": 0.3086,
+      "step": 6632
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002423785803237858,
+      "loss": 0.2896,
+      "step": 6633
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024236612702366127,
+      "loss": 0.2871,
+      "step": 6634
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002423536737235367,
+      "loss": 0.2362,
+      "step": 6635
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002423412204234122,
+      "loss": 0.2222,
+      "step": 6636
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024232876712328765,
+      "loss": 0.3822,
+      "step": 6637
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00024231631382316314,
+      "loss": 0.3834,
+      "step": 6638
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002423038605230386,
+      "loss": 0.2386,
+      "step": 6639
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024229140722291403,
+      "loss": 0.1872,
+      "step": 6640
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024227895392278951,
+      "loss": 0.2773,
+      "step": 6641
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024226650062266497,
+      "loss": 0.3194,
+      "step": 6642
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024225404732254046,
+      "loss": 0.4425,
+      "step": 6643
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024224159402241592,
+      "loss": 0.477,
+      "step": 6644
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002422291407222914,
+      "loss": 0.1842,
+      "step": 6645
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024221668742216684,
+      "loss": 0.2733,
+      "step": 6646
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024220423412204232,
+      "loss": 0.2812,
+      "step": 6647
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024219178082191778,
+      "loss": 0.2262,
+      "step": 6648
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024217932752179327,
+      "loss": 0.519,
+      "step": 6649
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024216687422166873,
+      "loss": 0.522,
+      "step": 6650
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002421544209215442,
+      "loss": 0.2566,
+      "step": 6651
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024214196762141965,
+      "loss": 0.2851,
+      "step": 6652
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024212951432129513,
+      "loss": 0.2757,
+      "step": 6653
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002421170610211706,
+      "loss": 0.2578,
+      "step": 6654
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024210460772104605,
+      "loss": 0.1936,
+      "step": 6655
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024209215442092154,
+      "loss": 0.3311,
+      "step": 6656
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024207970112079697,
+      "loss": 0.2347,
+      "step": 6657
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024206724782067246,
+      "loss": 0.2218,
+      "step": 6658
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024205479452054792,
+      "loss": 0.3063,
+      "step": 6659
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002420423412204234,
+      "loss": 0.2238,
+      "step": 6660
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024202988792029886,
+      "loss": 0.2114,
+      "step": 6661
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024201743462017432,
+      "loss": 0.303,
+      "step": 6662
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024200498132004978,
+      "loss": 0.2589,
+      "step": 6663
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024199252801992527,
+      "loss": 0.2442,
+      "step": 6664
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00024198007471980072,
+      "loss": 0.2133,
+      "step": 6665
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002419676214196762,
+      "loss": 0.2131,
+      "step": 6666
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024195516811955167,
+      "loss": 0.2367,
+      "step": 6667
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024194271481942713,
+      "loss": 0.549,
+      "step": 6668
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002419302615193026,
+      "loss": 0.3785,
+      "step": 6669
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024191780821917807,
+      "loss": 0.2278,
+      "step": 6670
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024190535491905353,
+      "loss": 0.2357,
+      "step": 6671
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000241892901618929,
+      "loss": 0.2391,
+      "step": 6672
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024188044831880445,
+      "loss": 0.2072,
+      "step": 6673
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002418679950186799,
+      "loss": 0.2215,
+      "step": 6674
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002418555417185554,
+      "loss": 0.4796,
+      "step": 6675
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024184308841843086,
+      "loss": 0.2093,
+      "step": 6676
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024183063511830634,
+      "loss": 0.2002,
+      "step": 6677
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002418181818181818,
+      "loss": 0.2845,
+      "step": 6678
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024180572851805726,
+      "loss": 0.2334,
+      "step": 6679
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024179327521793272,
+      "loss": 0.2517,
+      "step": 6680
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002417808219178082,
+      "loss": 0.4145,
+      "step": 6681
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024176836861768367,
+      "loss": 0.249,
+      "step": 6682
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024175591531755915,
+      "loss": 0.2329,
+      "step": 6683
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024174346201743458,
+      "loss": 0.1932,
+      "step": 6684
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024173100871731007,
+      "loss": 0.1904,
+      "step": 6685
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024171855541718553,
+      "loss": 0.1769,
+      "step": 6686
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000241706102117061,
+      "loss": 0.4565,
+      "step": 6687
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024169364881693647,
+      "loss": 0.2171,
+      "step": 6688
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024168119551681193,
+      "loss": 0.4491,
+      "step": 6689
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002416687422166874,
+      "loss": 0.1406,
+      "step": 6690
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024165628891656285,
+      "loss": 0.2542,
+      "step": 6691
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00024164383561643834,
+      "loss": 0.2368,
+      "step": 6692
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002416313823163138,
+      "loss": 0.1982,
+      "step": 6693
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024161892901618928,
+      "loss": 0.2668,
+      "step": 6694
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024160647571606472,
+      "loss": 0.3152,
+      "step": 6695
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002415940224159402,
+      "loss": 0.3469,
+      "step": 6696
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024158156911581566,
+      "loss": 0.1686,
+      "step": 6697
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024156911581569115,
+      "loss": 0.2478,
+      "step": 6698
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002415566625155666,
+      "loss": 0.2097,
+      "step": 6699
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002415442092154421,
+      "loss": 0.3619,
+      "step": 6700
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024153175591531753,
+      "loss": 0.229,
+      "step": 6701
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024151930261519298,
+      "loss": 0.2173,
+      "step": 6702
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024150684931506847,
+      "loss": 0.2303,
+      "step": 6703
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024149439601494393,
+      "loss": 0.2425,
+      "step": 6704
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024148194271481942,
+      "loss": 0.3031,
+      "step": 6705
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024146948941469485,
+      "loss": 0.4226,
+      "step": 6706
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024145703611457033,
+      "loss": 0.2196,
+      "step": 6707
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002414445828144458,
+      "loss": 0.2201,
+      "step": 6708
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024143212951432128,
+      "loss": 0.2012,
+      "step": 6709
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024141967621419674,
+      "loss": 0.1668,
+      "step": 6710
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024140722291407223,
+      "loss": 0.2553,
+      "step": 6711
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024139476961394766,
+      "loss": 0.5088,
+      "step": 6712
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024138231631382314,
+      "loss": 0.2256,
+      "step": 6713
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002413698630136986,
+      "loss": 0.2685,
+      "step": 6714
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002413574097135741,
+      "loss": 0.2352,
+      "step": 6715
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024134495641344955,
+      "loss": 0.2359,
+      "step": 6716
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024133250311332503,
+      "loss": 0.192,
+      "step": 6717
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00024132004981320047,
+      "loss": 0.5501,
+      "step": 6718
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024130759651307593,
+      "loss": 0.2075,
+      "step": 6719
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002412951432129514,
+      "loss": 0.2104,
+      "step": 6720
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024128268991282687,
+      "loss": 0.314,
+      "step": 6721
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024127023661270236,
+      "loss": 0.2258,
+      "step": 6722
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002412577833125778,
+      "loss": 0.2473,
+      "step": 6723
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024124533001245328,
+      "loss": 0.2989,
+      "step": 6724
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024123287671232874,
+      "loss": 0.4456,
+      "step": 6725
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024122042341220422,
+      "loss": 0.2501,
+      "step": 6726
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024120797011207968,
+      "loss": 0.2008,
+      "step": 6727
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024119551681195517,
+      "loss": 0.1945,
+      "step": 6728
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002411830635118306,
+      "loss": 0.2946,
+      "step": 6729
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024117061021170609,
+      "loss": 0.3595,
+      "step": 6730
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024115815691158154,
+      "loss": 0.3899,
+      "step": 6731
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024114570361145703,
+      "loss": 0.2336,
+      "step": 6732
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002411332503113325,
+      "loss": 0.2154,
+      "step": 6733
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024112079701120792,
+      "loss": 0.1922,
+      "step": 6734
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002411083437110834,
+      "loss": 0.1959,
+      "step": 6735
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024109589041095887,
+      "loss": 0.2579,
+      "step": 6736
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024108343711083435,
+      "loss": 0.4641,
+      "step": 6737
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002410709838107098,
+      "loss": 0.2351,
+      "step": 6738
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002410585305105853,
+      "loss": 0.2001,
+      "step": 6739
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024104607721046073,
+      "loss": 0.3623,
+      "step": 6740
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024103362391033622,
+      "loss": 0.1914,
+      "step": 6741
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024102117061021168,
+      "loss": 0.4364,
+      "step": 6742
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024100871731008716,
+      "loss": 0.4915,
+      "step": 6743
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00024099626400996262,
+      "loss": 0.2886,
+      "step": 6744
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002409838107098381,
+      "loss": 0.5409,
+      "step": 6745
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024097135740971354,
+      "loss": 0.2544,
+      "step": 6746
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024095890410958903,
+      "loss": 0.2149,
+      "step": 6747
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024094645080946449,
+      "loss": 0.2407,
+      "step": 6748
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024093399750933995,
+      "loss": 0.3197,
+      "step": 6749
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024092154420921543,
+      "loss": 0.5222,
+      "step": 6750
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024090909090909086,
+      "loss": 0.3118,
+      "step": 6751
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024089663760896635,
+      "loss": 0.2634,
+      "step": 6752
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002408841843088418,
+      "loss": 0.3509,
+      "step": 6753
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002408717310087173,
+      "loss": 0.2904,
+      "step": 6754
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024085927770859275,
+      "loss": 0.3218,
+      "step": 6755
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024084682440846824,
+      "loss": 0.4398,
+      "step": 6756
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024083437110834367,
+      "loss": 0.229,
+      "step": 6757
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024082191780821916,
+      "loss": 0.2507,
+      "step": 6758
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024080946450809462,
+      "loss": 0.2809,
+      "step": 6759
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002407970112079701,
+      "loss": 0.4981,
+      "step": 6760
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024078455790784556,
+      "loss": 0.1856,
+      "step": 6761
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024077210460772105,
+      "loss": 0.414,
+      "step": 6762
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024075965130759648,
+      "loss": 0.2489,
+      "step": 6763
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024074719800747197,
+      "loss": 0.3491,
+      "step": 6764
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024073474470734743,
+      "loss": 0.1861,
+      "step": 6765
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024072229140722289,
+      "loss": 0.3194,
+      "step": 6766
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024070983810709837,
+      "loss": 0.2873,
+      "step": 6767
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002406973848069738,
+      "loss": 0.3472,
+      "step": 6768
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0002406849315068493,
+      "loss": 0.2827,
+      "step": 6769
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00024067247820672475,
+      "loss": 0.3684,
+      "step": 6770
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024066002490660024,
+      "loss": 0.251,
+      "step": 6771
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0002406475716064757,
+      "loss": 0.2551,
+      "step": 6772
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024063511830635118,
+      "loss": 0.323,
+      "step": 6773
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024062266500622661,
+      "loss": 0.2574,
+      "step": 6774
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0002406102117061021,
+      "loss": 0.381,
+      "step": 6775
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024059775840597756,
+      "loss": 0.2274,
+      "step": 6776
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024058530510585305,
+      "loss": 0.2372,
+      "step": 6777
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0002405728518057285,
+      "loss": 0.2215,
+      "step": 6778
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000240560398505604,
+      "loss": 0.2522,
+      "step": 6779
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024054794520547942,
+      "loss": 0.2871,
+      "step": 6780
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024053549190535488,
+      "loss": 0.5783,
+      "step": 6781
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024052303860523037,
+      "loss": 0.293,
+      "step": 6782
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024051058530510583,
+      "loss": 0.2431,
+      "step": 6783
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024049813200498131,
+      "loss": 0.2168,
+      "step": 6784
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024048567870485675,
+      "loss": 0.1707,
+      "step": 6785
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024047322540473223,
+      "loss": 0.2805,
+      "step": 6786
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0002404607721046077,
+      "loss": 0.2476,
+      "step": 6787
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024044831880448318,
+      "loss": 0.2684,
+      "step": 6788
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024043586550435864,
+      "loss": 0.2525,
+      "step": 6789
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024042341220423412,
+      "loss": 0.2976,
+      "step": 6790
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024041095890410956,
+      "loss": 0.3447,
+      "step": 6791
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024039850560398504,
+      "loss": 0.187,
+      "step": 6792
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0002403860523038605,
+      "loss": 0.308,
+      "step": 6793
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000240373599003736,
+      "loss": 0.3007,
+      "step": 6794
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024036114570361145,
+      "loss": 0.1873,
+      "step": 6795
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00024034869240348688,
+      "loss": 0.1933,
+      "step": 6796
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024033623910336236,
+      "loss": 0.262,
+      "step": 6797
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024032378580323782,
+      "loss": 0.2663,
+      "step": 6798
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002403113325031133,
+      "loss": 0.305,
+      "step": 6799
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024029887920298877,
+      "loss": 0.5714,
+      "step": 6800
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024028642590286426,
+      "loss": 0.3078,
+      "step": 6801
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002402739726027397,
+      "loss": 0.2114,
+      "step": 6802
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024026151930261517,
+      "loss": 0.3006,
+      "step": 6803
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024024906600249063,
+      "loss": 0.2814,
+      "step": 6804
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024023661270236612,
+      "loss": 0.288,
+      "step": 6805
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024022415940224158,
+      "loss": 0.3677,
+      "step": 6806
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024021170610211706,
+      "loss": 0.2614,
+      "step": 6807
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002401992528019925,
+      "loss": 0.2602,
+      "step": 6808
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024018679950186798,
+      "loss": 0.3422,
+      "step": 6809
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024017434620174344,
+      "loss": 0.2387,
+      "step": 6810
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024016189290161893,
+      "loss": 0.2608,
+      "step": 6811
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002401494396014944,
+      "loss": 0.4565,
+      "step": 6812
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024013698630136982,
+      "loss": 0.2459,
+      "step": 6813
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002401245330012453,
+      "loss": 0.2315,
+      "step": 6814
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024011207970112077,
+      "loss": 0.2766,
+      "step": 6815
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024009962640099625,
+      "loss": 0.2738,
+      "step": 6816
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002400871731008717,
+      "loss": 0.1964,
+      "step": 6817
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002400747198007472,
+      "loss": 0.34,
+      "step": 6818
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024006226650062263,
+      "loss": 0.3153,
+      "step": 6819
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024004981320049812,
+      "loss": 0.1921,
+      "step": 6820
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024003735990037357,
+      "loss": 0.1645,
+      "step": 6821
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00024002490660024906,
+      "loss": 0.192,
+      "step": 6822
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00024001245330012452,
+      "loss": 0.2191,
+      "step": 6823
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.3844,
+      "step": 6824
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023998754669987544,
+      "loss": 0.6229,
+      "step": 6825
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023997509339975092,
+      "loss": 0.2266,
+      "step": 6826
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023996264009962638,
+      "loss": 0.1735,
+      "step": 6827
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023995018679950184,
+      "loss": 0.241,
+      "step": 6828
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023993773349937733,
+      "loss": 0.5057,
+      "step": 6829
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023992528019925276,
+      "loss": 0.1968,
+      "step": 6830
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023991282689912825,
+      "loss": 0.5151,
+      "step": 6831
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002399003735990037,
+      "loss": 0.2929,
+      "step": 6832
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002398879202988792,
+      "loss": 0.2938,
+      "step": 6833
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023987546699875465,
+      "loss": 0.2165,
+      "step": 6834
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002398630136986301,
+      "loss": 0.3973,
+      "step": 6835
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023985056039850557,
+      "loss": 0.3944,
+      "step": 6836
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023983810709838106,
+      "loss": 0.4293,
+      "step": 6837
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023982565379825652,
+      "loss": 0.6631,
+      "step": 6838
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000239813200498132,
+      "loss": 0.2565,
+      "step": 6839
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023980074719800746,
+      "loss": 0.4069,
+      "step": 6840
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023978829389788292,
+      "loss": 0.2223,
+      "step": 6841
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023977584059775838,
+      "loss": 0.2145,
+      "step": 6842
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023976338729763384,
+      "loss": 0.404,
+      "step": 6843
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023975093399750932,
+      "loss": 0.36,
+      "step": 6844
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023973848069738478,
+      "loss": 0.2587,
+      "step": 6845
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00023972602739726024,
+      "loss": 0.2294,
+      "step": 6846
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002397135740971357,
+      "loss": 0.2485,
+      "step": 6847
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002397011207970112,
+      "loss": 0.2152,
+      "step": 6848
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023968866749688665,
+      "loss": 0.2594,
+      "step": 6849
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023967621419676213,
+      "loss": 0.5414,
+      "step": 6850
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002396637608966376,
+      "loss": 0.2577,
+      "step": 6851
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023965130759651305,
+      "loss": 0.3396,
+      "step": 6852
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002396388542963885,
+      "loss": 0.2075,
+      "step": 6853
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000239626400996264,
+      "loss": 0.2886,
+      "step": 6854
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023961394769613946,
+      "loss": 0.2244,
+      "step": 6855
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023960149439601494,
+      "loss": 0.3208,
+      "step": 6856
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023958904109589038,
+      "loss": 0.267,
+      "step": 6857
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023957658779576586,
+      "loss": 0.2431,
+      "step": 6858
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023956413449564132,
+      "loss": 0.2107,
+      "step": 6859
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023955168119551678,
+      "loss": 0.1982,
+      "step": 6860
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023953922789539227,
+      "loss": 0.2998,
+      "step": 6861
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002395267745952677,
+      "loss": 0.3901,
+      "step": 6862
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023951432129514318,
+      "loss": 0.2594,
+      "step": 6863
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023950186799501864,
+      "loss": 0.2041,
+      "step": 6864
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023948941469489413,
+      "loss": 0.216,
+      "step": 6865
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002394769613947696,
+      "loss": 0.2739,
+      "step": 6866
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023946450809464508,
+      "loss": 0.2176,
+      "step": 6867
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002394520547945205,
+      "loss": 0.4097,
+      "step": 6868
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000239439601494396,
+      "loss": 0.2898,
+      "step": 6869
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023942714819427145,
+      "loss": 0.2147,
+      "step": 6870
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023941469489414694,
+      "loss": 0.1817,
+      "step": 6871
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002394022415940224,
+      "loss": 0.184,
+      "step": 6872
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023938978829389788,
+      "loss": 0.2031,
+      "step": 6873
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00023937733499377332,
+      "loss": 0.3059,
+      "step": 6874
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023936488169364878,
+      "loss": 0.3239,
+      "step": 6875
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023935242839352426,
+      "loss": 0.2799,
+      "step": 6876
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023933997509339972,
+      "loss": 0.4435,
+      "step": 6877
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0002393275217932752,
+      "loss": 0.2781,
+      "step": 6878
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023931506849315064,
+      "loss": 0.1973,
+      "step": 6879
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023930261519302613,
+      "loss": 0.2342,
+      "step": 6880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023929016189290159,
+      "loss": 0.4522,
+      "step": 6881
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023927770859277707,
+      "loss": 0.3336,
+      "step": 6882
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023926525529265253,
+      "loss": 0.2404,
+      "step": 6883
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023925280199252802,
+      "loss": 0.3068,
+      "step": 6884
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023924034869240345,
+      "loss": 0.1847,
+      "step": 6885
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023922789539227894,
+      "loss": 0.315,
+      "step": 6886
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0002392154420921544,
+      "loss": 0.3557,
+      "step": 6887
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023920298879202988,
+      "loss": 0.268,
+      "step": 6888
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023919053549190534,
+      "loss": 0.1897,
+      "step": 6889
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023917808219178077,
+      "loss": 0.1492,
+      "step": 6890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023916562889165626,
+      "loss": 0.2047,
+      "step": 6891
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023915317559153172,
+      "loss": 0.3225,
+      "step": 6892
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0002391407222914072,
+      "loss": 0.3591,
+      "step": 6893
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023912826899128266,
+      "loss": 0.2292,
+      "step": 6894
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023911581569115815,
+      "loss": 0.2181,
+      "step": 6895
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023910336239103358,
+      "loss": 0.2064,
+      "step": 6896
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023909090909090907,
+      "loss": 0.285,
+      "step": 6897
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023907845579078453,
+      "loss": 0.1983,
+      "step": 6898
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023906600249066,
+      "loss": 0.316,
+      "step": 6899
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00023905354919053547,
+      "loss": 0.3274,
+      "step": 6900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023904109589041096,
+      "loss": 0.2536,
+      "step": 6901
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002390286425902864,
+      "loss": 0.3054,
+      "step": 6902
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023901618929016188,
+      "loss": 0.1729,
+      "step": 6903
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023900373599003734,
+      "loss": 0.3311,
+      "step": 6904
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023899128268991282,
+      "loss": 0.3132,
+      "step": 6905
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023897882938978828,
+      "loss": 0.3945,
+      "step": 6906
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002389663760896637,
+      "loss": 0.299,
+      "step": 6907
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002389539227895392,
+      "loss": 0.2418,
+      "step": 6908
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023894146948941466,
+      "loss": 0.2054,
+      "step": 6909
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023892901618929015,
+      "loss": 0.232,
+      "step": 6910
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002389165628891656,
+      "loss": 0.2399,
+      "step": 6911
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002389041095890411,
+      "loss": 0.4857,
+      "step": 6912
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023889165628891652,
+      "loss": 0.2021,
+      "step": 6913
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.000238879202988792,
+      "loss": 0.2869,
+      "step": 6914
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023886674968866747,
+      "loss": 0.3331,
+      "step": 6915
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023885429638854295,
+      "loss": 0.2462,
+      "step": 6916
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002388418430884184,
+      "loss": 0.2731,
+      "step": 6917
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002388293897882939,
+      "loss": 0.507,
+      "step": 6918
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023881693648816933,
+      "loss": 0.2098,
+      "step": 6919
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023880448318804482,
+      "loss": 0.2618,
+      "step": 6920
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023879202988792028,
+      "loss": 0.4104,
+      "step": 6921
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023877957658779574,
+      "loss": 0.2085,
+      "step": 6922
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023876712328767122,
+      "loss": 0.2557,
+      "step": 6923
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023875466998754665,
+      "loss": 0.4392,
+      "step": 6924
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00023874221668742214,
+      "loss": 0.5811,
+      "step": 6925
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002387297633872976,
+      "loss": 0.201,
+      "step": 6926
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0002387173100871731,
+      "loss": 0.2776,
+      "step": 6927
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023870485678704855,
+      "loss": 0.3135,
+      "step": 6928
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023869240348692403,
+      "loss": 0.2834,
+      "step": 6929
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023867995018679946,
+      "loss": 0.3481,
+      "step": 6930
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023866749688667495,
+      "loss": 0.3544,
+      "step": 6931
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0002386550435865504,
+      "loss": 0.3024,
+      "step": 6932
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0002386425902864259,
+      "loss": 0.2357,
+      "step": 6933
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023863013698630135,
+      "loss": 0.354,
+      "step": 6934
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023861768368617684,
+      "loss": 0.2201,
+      "step": 6935
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023860523038605227,
+      "loss": 0.3222,
+      "step": 6936
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023859277708592776,
+      "loss": 0.4685,
+      "step": 6937
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023858032378580322,
+      "loss": 0.2858,
+      "step": 6938
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023856787048567868,
+      "loss": 0.2228,
+      "step": 6939
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023855541718555416,
+      "loss": 0.2347,
+      "step": 6940
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0002385429638854296,
+      "loss": 0.2214,
+      "step": 6941
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023853051058530508,
+      "loss": 0.2192,
+      "step": 6942
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023851805728518054,
+      "loss": 0.4102,
+      "step": 6943
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023850560398505603,
+      "loss": 0.2376,
+      "step": 6944
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0002384931506849315,
+      "loss": 0.2909,
+      "step": 6945
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023848069738480697,
+      "loss": 0.2443,
+      "step": 6946
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0002384682440846824,
+      "loss": 0.2137,
+      "step": 6947
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0002384557907845579,
+      "loss": 0.3174,
+      "step": 6948
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023844333748443335,
+      "loss": 0.5166,
+      "step": 6949
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023843088418430884,
+      "loss": 0.3905,
+      "step": 6950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0002384184308841843,
+      "loss": 0.2149,
+      "step": 6951
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00023840597758405978,
+      "loss": 0.3012,
+      "step": 6952
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023839352428393521,
+      "loss": 0.2104,
+      "step": 6953
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023838107098381067,
+      "loss": 0.3131,
+      "step": 6954
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023836861768368616,
+      "loss": 0.2611,
+      "step": 6955
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023835616438356162,
+      "loss": 0.4958,
+      "step": 6956
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0002383437110834371,
+      "loss": 0.4444,
+      "step": 6957
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023833125778331254,
+      "loss": 0.3826,
+      "step": 6958
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023831880448318802,
+      "loss": 0.2205,
+      "step": 6959
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023830635118306348,
+      "loss": 0.2049,
+      "step": 6960
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023829389788293897,
+      "loss": 0.2713,
+      "step": 6961
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023828144458281443,
+      "loss": 0.3698,
+      "step": 6962
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023826899128268991,
+      "loss": 0.3024,
+      "step": 6963
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023825653798256535,
+      "loss": 0.3025,
+      "step": 6964
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023824408468244083,
+      "loss": 0.2165,
+      "step": 6965
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0002382316313823163,
+      "loss": 0.2949,
+      "step": 6966
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023821917808219178,
+      "loss": 0.2697,
+      "step": 6967
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023820672478206724,
+      "loss": 0.4619,
+      "step": 6968
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023819427148194267,
+      "loss": 0.2802,
+      "step": 6969
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023818181818181816,
+      "loss": 0.2536,
+      "step": 6970
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023816936488169362,
+      "loss": 0.1974,
+      "step": 6971
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0002381569115815691,
+      "loss": 0.2088,
+      "step": 6972
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023814445828144456,
+      "loss": 0.2535,
+      "step": 6973
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023813200498132005,
+      "loss": 0.3417,
+      "step": 6974
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023811955168119548,
+      "loss": 0.315,
+      "step": 6975
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023810709838107097,
+      "loss": 0.2304,
+      "step": 6976
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023809464508094642,
+      "loss": 0.2819,
+      "step": 6977
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0002380821917808219,
+      "loss": 0.5372,
+      "step": 6978
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00023806973848069737,
+      "loss": 0.2307,
+      "step": 6979
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023805728518057286,
+      "loss": 0.3866,
+      "step": 6980
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002380448318804483,
+      "loss": 0.3,
+      "step": 6981
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023803237858032377,
+      "loss": 0.5023,
+      "step": 6982
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023801992528019923,
+      "loss": 0.279,
+      "step": 6983
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023800747198007472,
+      "loss": 0.2782,
+      "step": 6984
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023799501867995018,
+      "loss": 0.2418,
+      "step": 6985
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002379825653798256,
+      "loss": 0.3061,
+      "step": 6986
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002379701120797011,
+      "loss": 0.3754,
+      "step": 6987
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023795765877957656,
+      "loss": 0.3752,
+      "step": 6988
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023794520547945204,
+      "loss": 0.3836,
+      "step": 6989
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002379327521793275,
+      "loss": 0.2022,
+      "step": 6990
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000237920298879203,
+      "loss": 0.3669,
+      "step": 6991
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023790784557907842,
+      "loss": 0.2762,
+      "step": 6992
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002378953922789539,
+      "loss": 0.4257,
+      "step": 6993
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023788293897882937,
+      "loss": 0.3081,
+      "step": 6994
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023787048567870485,
+      "loss": 0.2645,
+      "step": 6995
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002378580323785803,
+      "loss": 0.3602,
+      "step": 6996
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023784557907845577,
+      "loss": 0.1781,
+      "step": 6997
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023783312577833123,
+      "loss": 0.2181,
+      "step": 6998
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023782067247820672,
+      "loss": 0.4099,
+      "step": 6999
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023780821917808218,
+      "loss": 0.3441,
+      "step": 7000
+    },
+    {
+      "epoch": 2.68,
+      "eval_cer": 0.11028737538287453,
+      "eval_loss": 0.37640687823295593,
+      "eval_runtime": 47.312,
+      "eval_samples_per_second": 37.052,
+      "eval_steps_per_second": 4.65,
+      "eval_wer": 0.18873602751504728,
+      "step": 7000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023779576587795763,
+      "loss": 0.3306,
+      "step": 7001
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023778331257783312,
+      "loss": 0.1705,
+      "step": 7002
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023777085927770855,
+      "loss": 0.1954,
+      "step": 7003
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00023775840597758404,
+      "loss": 0.255,
+      "step": 7004
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002377459526774595,
+      "loss": 0.5178,
+      "step": 7005
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023773349937733498,
+      "loss": 0.2846,
+      "step": 7006
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023772104607721044,
+      "loss": 0.4779,
+      "step": 7007
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002377085927770859,
+      "loss": 0.2885,
+      "step": 7008
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023769613947696136,
+      "loss": 0.3564,
+      "step": 7009
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023768368617683685,
+      "loss": 0.1786,
+      "step": 7010
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002376712328767123,
+      "loss": 0.2676,
+      "step": 7011
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002376587795765878,
+      "loss": 0.4086,
+      "step": 7012
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023764632627646325,
+      "loss": 0.3416,
+      "step": 7013
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002376338729763387,
+      "loss": 0.2596,
+      "step": 7014
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023762141967621417,
+      "loss": 0.1977,
+      "step": 7015
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023760896637608963,
+      "loss": 0.2189,
+      "step": 7016
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023759651307596512,
+      "loss": 0.2242,
+      "step": 7017
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023758405977584058,
+      "loss": 0.6218,
+      "step": 7018
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023757160647571603,
+      "loss": 0.2905,
+      "step": 7019
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002375591531755915,
+      "loss": 0.2143,
+      "step": 7020
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023754669987546698,
+      "loss": 0.1523,
+      "step": 7021
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023753424657534244,
+      "loss": 0.1833,
+      "step": 7022
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023752179327521793,
+      "loss": 0.359,
+      "step": 7023
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023750933997509336,
+      "loss": 0.2912,
+      "step": 7024
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023749688667496884,
+      "loss": 0.3056,
+      "step": 7025
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002374844333748443,
+      "loss": 0.7801,
+      "step": 7026
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002374719800747198,
+      "loss": 0.336,
+      "step": 7027
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023745952677459525,
+      "loss": 0.4029,
+      "step": 7028
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023744707347447073,
+      "loss": 0.3624,
+      "step": 7029
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023743462017434617,
+      "loss": 0.3636,
+      "step": 7030
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00023742216687422165,
+      "loss": 0.5806,
+      "step": 7031
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0002374097135740971,
+      "loss": 0.279,
+      "step": 7032
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023739726027397257,
+      "loss": 0.252,
+      "step": 7033
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023738480697384806,
+      "loss": 0.2839,
+      "step": 7034
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0002373723536737235,
+      "loss": 0.2362,
+      "step": 7035
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023735990037359898,
+      "loss": 0.2523,
+      "step": 7036
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023734744707347444,
+      "loss": 0.408,
+      "step": 7037
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023733499377334992,
+      "loss": 0.3133,
+      "step": 7038
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023732254047322538,
+      "loss": 0.4108,
+      "step": 7039
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023731008717310087,
+      "loss": 0.1871,
+      "step": 7040
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0002372976338729763,
+      "loss": 0.3397,
+      "step": 7041
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023728518057285179,
+      "loss": 0.2989,
+      "step": 7042
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023727272727272724,
+      "loss": 0.4117,
+      "step": 7043
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023726027397260273,
+      "loss": 0.47,
+      "step": 7044
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0002372478206724782,
+      "loss": 0.2329,
+      "step": 7045
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023723536737235368,
+      "loss": 0.2166,
+      "step": 7046
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0002372229140722291,
+      "loss": 0.2732,
+      "step": 7047
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023721046077210457,
+      "loss": 0.3816,
+      "step": 7048
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023719800747198005,
+      "loss": 0.37,
+      "step": 7049
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0002371855541718555,
+      "loss": 0.4837,
+      "step": 7050
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000237173100871731,
+      "loss": 0.2115,
+      "step": 7051
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023716064757160643,
+      "loss": 0.1933,
+      "step": 7052
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023714819427148192,
+      "loss": 0.2487,
+      "step": 7053
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023713574097135738,
+      "loss": 0.2214,
+      "step": 7054
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023712328767123286,
+      "loss": 0.324,
+      "step": 7055
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00023711083437110832,
+      "loss": 0.4278,
+      "step": 7056
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0002370983810709838,
+      "loss": 0.3052,
+      "step": 7057
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023708592777085924,
+      "loss": 0.2248,
+      "step": 7058
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023707347447073473,
+      "loss": 0.2086,
+      "step": 7059
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023706102117061019,
+      "loss": 0.2273,
+      "step": 7060
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023704856787048567,
+      "loss": 0.2381,
+      "step": 7061
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023703611457036113,
+      "loss": 0.4194,
+      "step": 7062
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023702366127023656,
+      "loss": 0.3798,
+      "step": 7063
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023701120797011205,
+      "loss": 0.4211,
+      "step": 7064
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0002369987546699875,
+      "loss": 0.2727,
+      "step": 7065
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000236986301369863,
+      "loss": 0.2057,
+      "step": 7066
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023697384806973845,
+      "loss": 0.2669,
+      "step": 7067
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023696139476961394,
+      "loss": 0.5615,
+      "step": 7068
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023694894146948937,
+      "loss": 0.3654,
+      "step": 7069
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023693648816936486,
+      "loss": 0.3927,
+      "step": 7070
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023692403486924032,
+      "loss": 0.2701,
+      "step": 7071
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0002369115815691158,
+      "loss": 0.3159,
+      "step": 7072
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023689912826899126,
+      "loss": 0.267,
+      "step": 7073
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023688667496886675,
+      "loss": 0.4304,
+      "step": 7074
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023687422166874218,
+      "loss": 0.2849,
+      "step": 7075
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023686176836861767,
+      "loss": 0.2049,
+      "step": 7076
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023684931506849313,
+      "loss": 0.2649,
+      "step": 7077
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023683686176836861,
+      "loss": 0.2739,
+      "step": 7078
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023682440846824407,
+      "loss": 0.221,
+      "step": 7079
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0002368119551681195,
+      "loss": 0.3524,
+      "step": 7080
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000236799501867995,
+      "loss": 0.3192,
+      "step": 7081
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023678704856787045,
+      "loss": 0.2658,
+      "step": 7082
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00023677459526774594,
+      "loss": 0.251,
+      "step": 7083
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002367621419676214,
+      "loss": 0.4408,
+      "step": 7084
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023674968866749688,
+      "loss": 0.2355,
+      "step": 7085
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023673723536737231,
+      "loss": 0.3394,
+      "step": 7086
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002367247820672478,
+      "loss": 0.5107,
+      "step": 7087
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023671232876712326,
+      "loss": 0.3667,
+      "step": 7088
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023669987546699875,
+      "loss": 0.2239,
+      "step": 7089
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002366874221668742,
+      "loss": 0.2925,
+      "step": 7090
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002366749688667497,
+      "loss": 0.2461,
+      "step": 7091
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023666251556662512,
+      "loss": 0.2639,
+      "step": 7092
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002366500622665006,
+      "loss": 0.4315,
+      "step": 7093
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023663760896637607,
+      "loss": 0.2344,
+      "step": 7094
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023662515566625153,
+      "loss": 0.2179,
+      "step": 7095
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023661270236612701,
+      "loss": 0.2314,
+      "step": 7096
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023660024906600245,
+      "loss": 0.2536,
+      "step": 7097
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023658779576587793,
+      "loss": 0.3899,
+      "step": 7098
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002365753424657534,
+      "loss": 0.2887,
+      "step": 7099
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023656288916562888,
+      "loss": 0.4058,
+      "step": 7100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023655043586550434,
+      "loss": 0.3151,
+      "step": 7101
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023653798256537982,
+      "loss": 0.3253,
+      "step": 7102
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023652552926525526,
+      "loss": 0.1544,
+      "step": 7103
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023651307596513074,
+      "loss": 0.2911,
+      "step": 7104
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002365006226650062,
+      "loss": 0.2813,
+      "step": 7105
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002364881693648817,
+      "loss": 0.3582,
+      "step": 7106
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023647571606475715,
+      "loss": 0.3057,
+      "step": 7107
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023646326276463263,
+      "loss": 0.2099,
+      "step": 7108
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00023645080946450806,
+      "loss": 0.3405,
+      "step": 7109
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023643835616438352,
+      "loss": 0.3299,
+      "step": 7110
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000236425902864259,
+      "loss": 0.2218,
+      "step": 7111
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023641344956413447,
+      "loss": 0.4578,
+      "step": 7112
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023640099626400996,
+      "loss": 0.2431,
+      "step": 7113
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002363885429638854,
+      "loss": 0.2128,
+      "step": 7114
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023637608966376087,
+      "loss": 0.163,
+      "step": 7115
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023636363636363633,
+      "loss": 0.3055,
+      "step": 7116
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023635118306351182,
+      "loss": 0.2326,
+      "step": 7117
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023633872976338728,
+      "loss": 0.323,
+      "step": 7118
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023632627646326276,
+      "loss": 0.7871,
+      "step": 7119
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002363138231631382,
+      "loss": 0.3326,
+      "step": 7120
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023630136986301368,
+      "loss": 0.3115,
+      "step": 7121
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023628891656288914,
+      "loss": 0.2478,
+      "step": 7122
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023627646326276463,
+      "loss": 0.2188,
+      "step": 7123
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002362640099626401,
+      "loss": 0.2164,
+      "step": 7124
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023625155666251557,
+      "loss": 0.3267,
+      "step": 7125
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000236239103362391,
+      "loss": 0.3614,
+      "step": 7126
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023622665006226647,
+      "loss": 0.562,
+      "step": 7127
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023621419676214195,
+      "loss": 0.2687,
+      "step": 7128
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002362017434620174,
+      "loss": 0.2341,
+      "step": 7129
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002361892901618929,
+      "loss": 0.2372,
+      "step": 7130
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023617683686176833,
+      "loss": 0.44,
+      "step": 7131
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023616438356164382,
+      "loss": 0.2727,
+      "step": 7132
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023615193026151927,
+      "loss": 0.2701,
+      "step": 7133
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023613947696139476,
+      "loss": 0.3407,
+      "step": 7134
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00023612702366127022,
+      "loss": 0.1854,
+      "step": 7135
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002361145703611457,
+      "loss": 0.2655,
+      "step": 7136
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023610211706102114,
+      "loss": 0.313,
+      "step": 7137
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023608966376089662,
+      "loss": 0.2463,
+      "step": 7138
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023607721046077208,
+      "loss": 0.3118,
+      "step": 7139
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023606475716064757,
+      "loss": 0.2578,
+      "step": 7140
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023605230386052303,
+      "loss": 0.191,
+      "step": 7141
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023603985056039846,
+      "loss": 0.2306,
+      "step": 7142
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023602739726027395,
+      "loss": 0.395,
+      "step": 7143
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002360149439601494,
+      "loss": 0.3612,
+      "step": 7144
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002360024906600249,
+      "loss": 0.309,
+      "step": 7145
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023599003735990035,
+      "loss": 0.2498,
+      "step": 7146
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023597758405977584,
+      "loss": 0.2079,
+      "step": 7147
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023596513075965127,
+      "loss": 0.3225,
+      "step": 7148
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023595267745952676,
+      "loss": 0.321,
+      "step": 7149
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023594022415940222,
+      "loss": 0.3659,
+      "step": 7150
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002359277708592777,
+      "loss": 0.2514,
+      "step": 7151
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023591531755915316,
+      "loss": 0.2154,
+      "step": 7152
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023590286425902865,
+      "loss": 0.2146,
+      "step": 7153
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023589041095890408,
+      "loss": 0.1787,
+      "step": 7154
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023587795765877957,
+      "loss": 0.4282,
+      "step": 7155
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023586550435865503,
+      "loss": 0.4179,
+      "step": 7156
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023585305105853048,
+      "loss": 0.2656,
+      "step": 7157
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023584059775840597,
+      "loss": 0.2567,
+      "step": 7158
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002358281444582814,
+      "loss": 0.2381,
+      "step": 7159
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002358156911581569,
+      "loss": 0.3187,
+      "step": 7160
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00023580323785803235,
+      "loss": 0.2709,
+      "step": 7161
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023579078455790783,
+      "loss": 0.4439,
+      "step": 7162
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002357783312577833,
+      "loss": 0.2432,
+      "step": 7163
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023576587795765878,
+      "loss": 0.3034,
+      "step": 7164
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002357534246575342,
+      "loss": 0.3636,
+      "step": 7165
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002357409713574097,
+      "loss": 0.3608,
+      "step": 7166
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023572851805728516,
+      "loss": 0.2341,
+      "step": 7167
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023571606475716064,
+      "loss": 0.4733,
+      "step": 7168
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002357036114570361,
+      "loss": 0.2681,
+      "step": 7169
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023569115815691156,
+      "loss": 0.2044,
+      "step": 7170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023567870485678702,
+      "loss": 0.2605,
+      "step": 7171
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002356662515566625,
+      "loss": 0.2557,
+      "step": 7172
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023565379825653797,
+      "loss": 0.3147,
+      "step": 7173
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023564134495641343,
+      "loss": 0.3278,
+      "step": 7174
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002356288916562889,
+      "loss": 0.2889,
+      "step": 7175
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023561643835616434,
+      "loss": 0.248,
+      "step": 7176
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023560398505603983,
+      "loss": 0.183,
+      "step": 7177
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002355915317559153,
+      "loss": 0.2153,
+      "step": 7178
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023557907845579078,
+      "loss": 0.3695,
+      "step": 7179
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023556662515566623,
+      "loss": 0.2406,
+      "step": 7180
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002355541718555417,
+      "loss": 0.6405,
+      "step": 7181
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023554171855541715,
+      "loss": 0.2004,
+      "step": 7182
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023552926525529264,
+      "loss": 0.2345,
+      "step": 7183
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002355168119551681,
+      "loss": 0.1317,
+      "step": 7184
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023550435865504358,
+      "loss": 0.2263,
+      "step": 7185
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00023549190535491902,
+      "loss": 0.2268,
+      "step": 7186
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002354794520547945,
+      "loss": 0.4665,
+      "step": 7187
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023546699875466996,
+      "loss": 0.3013,
+      "step": 7188
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023545454545454542,
+      "loss": 0.2346,
+      "step": 7189
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002354420921544209,
+      "loss": 0.1679,
+      "step": 7190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023542963885429637,
+      "loss": 0.1944,
+      "step": 7191
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023541718555417183,
+      "loss": 0.2758,
+      "step": 7192
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023540473225404729,
+      "loss": 0.3748,
+      "step": 7193
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023539227895392277,
+      "loss": 0.2905,
+      "step": 7194
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023537982565379823,
+      "loss": 0.2266,
+      "step": 7195
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023536737235367372,
+      "loss": 0.182,
+      "step": 7196
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023535491905354915,
+      "loss": 0.2819,
+      "step": 7197
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023534246575342464,
+      "loss": 0.2601,
+      "step": 7198
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002353300124533001,
+      "loss": 0.3213,
+      "step": 7199
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023531755915317558,
+      "loss": 0.4133,
+      "step": 7200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023530510585305104,
+      "loss": 0.2084,
+      "step": 7201
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023529265255292653,
+      "loss": 0.2472,
+      "step": 7202
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023528019925280196,
+      "loss": 0.1645,
+      "step": 7203
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023526774595267744,
+      "loss": 0.2868,
+      "step": 7204
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002352552926525529,
+      "loss": 0.3558,
+      "step": 7205
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023524283935242836,
+      "loss": 0.3569,
+      "step": 7206
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023523038605230385,
+      "loss": 0.2267,
+      "step": 7207
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023521793275217928,
+      "loss": 0.187,
+      "step": 7208
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023520547945205477,
+      "loss": 0.219,
+      "step": 7209
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023519302615193023,
+      "loss": 0.2175,
+      "step": 7210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002351805728518057,
+      "loss": 0.1738,
+      "step": 7211
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023516811955168117,
+      "loss": 0.378,
+      "step": 7212
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00023515566625155666,
+      "loss": 0.396,
+      "step": 7213
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002351432129514321,
+      "loss": 0.2316,
+      "step": 7214
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023513075965130758,
+      "loss": 0.3112,
+      "step": 7215
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023511830635118304,
+      "loss": 0.2202,
+      "step": 7216
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023510585305105852,
+      "loss": 0.2509,
+      "step": 7217
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023509339975093398,
+      "loss": 0.4338,
+      "step": 7218
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023508094645080947,
+      "loss": 0.2562,
+      "step": 7219
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002350684931506849,
+      "loss": 0.2735,
+      "step": 7220
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023505603985056036,
+      "loss": 0.2877,
+      "step": 7221
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023504358655043585,
+      "loss": 0.2176,
+      "step": 7222
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002350311332503113,
+      "loss": 0.3294,
+      "step": 7223
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002350186799501868,
+      "loss": 0.3457,
+      "step": 7224
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023500622665006222,
+      "loss": 0.477,
+      "step": 7225
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002349937733499377,
+      "loss": 0.3802,
+      "step": 7226
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023498132004981317,
+      "loss": 0.2907,
+      "step": 7227
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023496886674968865,
+      "loss": 0.1982,
+      "step": 7228
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023495641344956411,
+      "loss": 0.2194,
+      "step": 7229
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002349439601494396,
+      "loss": 0.2193,
+      "step": 7230
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023493150684931503,
+      "loss": 0.3837,
+      "step": 7231
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023491905354919052,
+      "loss": 0.1957,
+      "step": 7232
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023490660024906598,
+      "loss": 0.4223,
+      "step": 7233
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023489414694894146,
+      "loss": 0.2165,
+      "step": 7234
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023488169364881692,
+      "loss": 0.2423,
+      "step": 7235
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023486924034869235,
+      "loss": 0.3875,
+      "step": 7236
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00023485678704856784,
+      "loss": 0.4214,
+      "step": 7237
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002348443337484433,
+      "loss": 0.2808,
+      "step": 7238
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002348318804483188,
+      "loss": 0.2756,
+      "step": 7239
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023481942714819425,
+      "loss": 0.2027,
+      "step": 7240
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023480697384806973,
+      "loss": 0.189,
+      "step": 7241
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023479452054794516,
+      "loss": 0.2118,
+      "step": 7242
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023478206724782065,
+      "loss": 0.4005,
+      "step": 7243
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002347696139476961,
+      "loss": 0.2499,
+      "step": 7244
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002347571606475716,
+      "loss": 0.2153,
+      "step": 7245
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023474470734744706,
+      "loss": 0.2071,
+      "step": 7246
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023473225404732254,
+      "loss": 0.3753,
+      "step": 7247
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023471980074719797,
+      "loss": 0.1938,
+      "step": 7248
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023470734744707346,
+      "loss": 0.2367,
+      "step": 7249
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023469489414694892,
+      "loss": 0.3874,
+      "step": 7250
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002346824408468244,
+      "loss": 0.2412,
+      "step": 7251
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023466998754669986,
+      "loss": 0.3173,
+      "step": 7252
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002346575342465753,
+      "loss": 0.4021,
+      "step": 7253
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023464508094645078,
+      "loss": 0.2075,
+      "step": 7254
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023463262764632624,
+      "loss": 0.303,
+      "step": 7255
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023462017434620173,
+      "loss": 0.3871,
+      "step": 7256
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002346077210460772,
+      "loss": 0.1823,
+      "step": 7257
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023459526774595267,
+      "loss": 0.2208,
+      "step": 7258
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002345828144458281,
+      "loss": 0.3008,
+      "step": 7259
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002345703611457036,
+      "loss": 0.2487,
+      "step": 7260
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023455790784557905,
+      "loss": 0.3231,
+      "step": 7261
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023454545454545454,
+      "loss": 0.3867,
+      "step": 7262
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023453300124533,
+      "loss": 0.2462,
+      "step": 7263
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023452054794520548,
+      "loss": 0.2787,
+      "step": 7264
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00023450809464508091,
+      "loss": 0.212,
+      "step": 7265
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002344956413449564,
+      "loss": 0.2315,
+      "step": 7266
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023448318804483186,
+      "loss": 0.2433,
+      "step": 7267
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023447073474470732,
+      "loss": 0.3966,
+      "step": 7268
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002344582814445828,
+      "loss": 0.2663,
+      "step": 7269
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023444582814445824,
+      "loss": 0.3166,
+      "step": 7270
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023443337484433372,
+      "loss": 0.3086,
+      "step": 7271
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023442092154420918,
+      "loss": 0.1944,
+      "step": 7272
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023440846824408467,
+      "loss": 0.183,
+      "step": 7273
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023439601494396013,
+      "loss": 0.2776,
+      "step": 7274
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023438356164383561,
+      "loss": 0.3432,
+      "step": 7275
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023437110834371105,
+      "loss": 0.2149,
+      "step": 7276
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023435865504358653,
+      "loss": 0.2295,
+      "step": 7277
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000234346201743462,
+      "loss": 0.1834,
+      "step": 7278
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023433374844333748,
+      "loss": 0.2356,
+      "step": 7279
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023432129514321294,
+      "loss": 0.2575,
+      "step": 7280
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023430884184308842,
+      "loss": 0.5368,
+      "step": 7281
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023429638854296386,
+      "loss": 0.7303,
+      "step": 7282
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023428393524283932,
+      "loss": 0.2344,
+      "step": 7283
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002342714819427148,
+      "loss": 0.3201,
+      "step": 7284
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023425902864259026,
+      "loss": 0.3512,
+      "step": 7285
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023424657534246575,
+      "loss": 0.3054,
+      "step": 7286
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023423412204234118,
+      "loss": 0.5612,
+      "step": 7287
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023422166874221667,
+      "loss": 0.3096,
+      "step": 7288
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023420921544209212,
+      "loss": 0.2806,
+      "step": 7289
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002341967621419676,
+      "loss": 0.1769,
+      "step": 7290
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023418430884184307,
+      "loss": 0.2434,
+      "step": 7291
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00023417185554171856,
+      "loss": 0.2239,
+      "step": 7292
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000234159402241594,
+      "loss": 0.2859,
+      "step": 7293
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023414694894146947,
+      "loss": 0.2918,
+      "step": 7294
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023413449564134493,
+      "loss": 0.2343,
+      "step": 7295
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023412204234122042,
+      "loss": 0.1857,
+      "step": 7296
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023410958904109588,
+      "loss": 0.1832,
+      "step": 7297
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023409713574097137,
+      "loss": 0.3734,
+      "step": 7298
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002340846824408468,
+      "loss": 0.2362,
+      "step": 7299
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023407222914072226,
+      "loss": 0.4165,
+      "step": 7300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023405977584059774,
+      "loss": 0.2267,
+      "step": 7301
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002340473225404732,
+      "loss": 0.2938,
+      "step": 7302
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002340348692403487,
+      "loss": 0.1857,
+      "step": 7303
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023402241594022412,
+      "loss": 0.2134,
+      "step": 7304
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002340099626400996,
+      "loss": 0.2701,
+      "step": 7305
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023399750933997507,
+      "loss": 0.386,
+      "step": 7306
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023398505603985055,
+      "loss": 0.2056,
+      "step": 7307
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000233972602739726,
+      "loss": 0.2993,
+      "step": 7308
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002339601494396015,
+      "loss": 0.2211,
+      "step": 7309
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023394769613947693,
+      "loss": 0.2383,
+      "step": 7310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023393524283935242,
+      "loss": 0.4231,
+      "step": 7311
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023392278953922788,
+      "loss": 0.3875,
+      "step": 7312
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023391033623910336,
+      "loss": 0.3763,
+      "step": 7313
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023389788293897882,
+      "loss": 0.2378,
+      "step": 7314
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023388542963885425,
+      "loss": 0.2028,
+      "step": 7315
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023387297633872974,
+      "loss": 0.2734,
+      "step": 7316
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002338605230386052,
+      "loss": 0.2338,
+      "step": 7317
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00023384806973848068,
+      "loss": 0.4062,
+      "step": 7318
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023383561643835614,
+      "loss": 0.2119,
+      "step": 7319
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023382316313823163,
+      "loss": 0.3443,
+      "step": 7320
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023381070983810706,
+      "loss": 0.197,
+      "step": 7321
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023379825653798255,
+      "loss": 0.2591,
+      "step": 7322
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000233785803237858,
+      "loss": 0.182,
+      "step": 7323
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002337733499377335,
+      "loss": 0.3646,
+      "step": 7324
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023376089663760895,
+      "loss": 0.3482,
+      "step": 7325
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023374844333748444,
+      "loss": 0.2402,
+      "step": 7326
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023373599003735987,
+      "loss": 0.2781,
+      "step": 7327
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023372353673723536,
+      "loss": 0.252,
+      "step": 7328
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023371108343711082,
+      "loss": 0.2101,
+      "step": 7329
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023369863013698628,
+      "loss": 0.2101,
+      "step": 7330
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023368617683686176,
+      "loss": 0.4848,
+      "step": 7331
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002336737235367372,
+      "loss": 0.2308,
+      "step": 7332
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023366127023661268,
+      "loss": 0.2588,
+      "step": 7333
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023364881693648814,
+      "loss": 0.1906,
+      "step": 7334
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023363636363636363,
+      "loss": 0.2007,
+      "step": 7335
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023362391033623908,
+      "loss": 0.3465,
+      "step": 7336
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023361145703611457,
+      "loss": 0.425,
+      "step": 7337
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023359900373599,
+      "loss": 0.2332,
+      "step": 7338
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002335865504358655,
+      "loss": 0.178,
+      "step": 7339
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023357409713574095,
+      "loss": 0.2345,
+      "step": 7340
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023356164383561643,
+      "loss": 0.2971,
+      "step": 7341
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002335491905354919,
+      "loss": 0.2043,
+      "step": 7342
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00023353673723536735,
+      "loss": 0.344,
+      "step": 7343
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002335242839352428,
+      "loss": 0.2399,
+      "step": 7344
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002335118306351183,
+      "loss": 0.2322,
+      "step": 7345
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023349937733499376,
+      "loss": 0.2264,
+      "step": 7346
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023348692403486922,
+      "loss": 0.201,
+      "step": 7347
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023347447073474468,
+      "loss": 0.2219,
+      "step": 7348
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023346201743462014,
+      "loss": 0.3194,
+      "step": 7349
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023344956413449562,
+      "loss": 0.4711,
+      "step": 7350
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023343711083437108,
+      "loss": 0.5584,
+      "step": 7351
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023342465753424657,
+      "loss": 0.4863,
+      "step": 7352
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023341220423412203,
+      "loss": 0.1866,
+      "step": 7353
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023339975093399749,
+      "loss": 0.2085,
+      "step": 7354
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023338729763387294,
+      "loss": 0.3231,
+      "step": 7355
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023337484433374843,
+      "loss": 0.4285,
+      "step": 7356
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002333623910336239,
+      "loss": 0.253,
+      "step": 7357
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023334993773349938,
+      "loss": 0.2781,
+      "step": 7358
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002333374844333748,
+      "loss": 0.2775,
+      "step": 7359
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002333250311332503,
+      "loss": 0.2959,
+      "step": 7360
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023331257783312575,
+      "loss": 0.3168,
+      "step": 7361
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002333001245330012,
+      "loss": 0.2867,
+      "step": 7362
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002332876712328767,
+      "loss": 0.2411,
+      "step": 7363
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023327521793275216,
+      "loss": 0.2376,
+      "step": 7364
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023326276463262762,
+      "loss": 0.2702,
+      "step": 7365
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023325031133250308,
+      "loss": 0.313,
+      "step": 7366
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023323785803237856,
+      "loss": 0.4111,
+      "step": 7367
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023322540473225402,
+      "loss": 0.3907,
+      "step": 7368
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002332129514321295,
+      "loss": 0.2625,
+      "step": 7369
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00023320049813200494,
+      "loss": 0.3091,
+      "step": 7370
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023318804483188043,
+      "loss": 0.2096,
+      "step": 7371
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023317559153175589,
+      "loss": 0.1935,
+      "step": 7372
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023316313823163137,
+      "loss": 0.2207,
+      "step": 7373
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023315068493150683,
+      "loss": 0.3323,
+      "step": 7374
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023313823163138232,
+      "loss": 0.5642,
+      "step": 7375
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023312577833125775,
+      "loss": 0.2105,
+      "step": 7376
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0002331133250311332,
+      "loss": 0.1957,
+      "step": 7377
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0002331008717310087,
+      "loss": 0.1578,
+      "step": 7378
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023308841843088415,
+      "loss": 0.2831,
+      "step": 7379
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023307596513075964,
+      "loss": 0.3989,
+      "step": 7380
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023306351183063507,
+      "loss": 0.4058,
+      "step": 7381
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023305105853051056,
+      "loss": 0.2988,
+      "step": 7382
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023303860523038602,
+      "loss": 0.4761,
+      "step": 7383
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0002330261519302615,
+      "loss": 0.1658,
+      "step": 7384
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023301369863013696,
+      "loss": 0.213,
+      "step": 7385
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023300124533001245,
+      "loss": 0.2553,
+      "step": 7386
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023298879202988788,
+      "loss": 0.4636,
+      "step": 7387
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023297633872976337,
+      "loss": 0.288,
+      "step": 7388
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023296388542963883,
+      "loss": 0.2177,
+      "step": 7389
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023295143212951431,
+      "loss": 0.2349,
+      "step": 7390
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023293897882938977,
+      "loss": 0.2187,
+      "step": 7391
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023292652552926526,
+      "loss": 0.1842,
+      "step": 7392
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0002329140722291407,
+      "loss": 0.3179,
+      "step": 7393
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023290161892901615,
+      "loss": 0.3198,
+      "step": 7394
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00023288916562889164,
+      "loss": 0.1661,
+      "step": 7395
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0002328767123287671,
+      "loss": 0.1802,
+      "step": 7396
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023286425902864258,
+      "loss": 0.161,
+      "step": 7397
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023285180572851801,
+      "loss": 0.2384,
+      "step": 7398
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002328393524283935,
+      "loss": 0.2841,
+      "step": 7399
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023282689912826896,
+      "loss": 0.3059,
+      "step": 7400
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023281444582814445,
+      "loss": 0.2697,
+      "step": 7401
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002328019925280199,
+      "loss": 0.1852,
+      "step": 7402
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002327895392278954,
+      "loss": 0.4483,
+      "step": 7403
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023277708592777082,
+      "loss": 0.243,
+      "step": 7404
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002327646326276463,
+      "loss": 0.28,
+      "step": 7405
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023275217932752177,
+      "loss": 0.6611,
+      "step": 7406
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023273972602739726,
+      "loss": 0.2098,
+      "step": 7407
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023272727272727271,
+      "loss": 0.2225,
+      "step": 7408
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023271481942714815,
+      "loss": 0.2683,
+      "step": 7409
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023270236612702363,
+      "loss": 0.2229,
+      "step": 7410
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002326899128268991,
+      "loss": 0.2025,
+      "step": 7411
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023267745952677458,
+      "loss": 0.5325,
+      "step": 7412
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023266500622665004,
+      "loss": 0.2645,
+      "step": 7413
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023265255292652552,
+      "loss": 0.383,
+      "step": 7414
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023264009962640096,
+      "loss": 0.2447,
+      "step": 7415
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023262764632627644,
+      "loss": 0.3079,
+      "step": 7416
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002326151930261519,
+      "loss": 0.3173,
+      "step": 7417
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002326027397260274,
+      "loss": 0.5291,
+      "step": 7418
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023259028642590285,
+      "loss": 0.4041,
+      "step": 7419
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023257783312577833,
+      "loss": 0.2436,
+      "step": 7420
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023256537982565376,
+      "loss": 0.2052,
+      "step": 7421
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00023255292652552925,
+      "loss": 0.3296,
+      "step": 7422
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002325404732254047,
+      "loss": 0.2304,
+      "step": 7423
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023252801992528017,
+      "loss": 0.3288,
+      "step": 7424
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023251556662515566,
+      "loss": 0.3121,
+      "step": 7425
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002325031133250311,
+      "loss": 0.2511,
+      "step": 7426
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023249066002490657,
+      "loss": 0.2065,
+      "step": 7427
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023247820672478203,
+      "loss": 0.2059,
+      "step": 7428
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023246575342465752,
+      "loss": 0.2345,
+      "step": 7429
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023245330012453298,
+      "loss": 0.2825,
+      "step": 7430
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023244084682440846,
+      "loss": 0.3911,
+      "step": 7431
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002324283935242839,
+      "loss": 0.3568,
+      "step": 7432
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023241594022415938,
+      "loss": 0.1997,
+      "step": 7433
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023240348692403484,
+      "loss": 0.248,
+      "step": 7434
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023239103362391033,
+      "loss": 0.2464,
+      "step": 7435
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002323785803237858,
+      "loss": 0.1874,
+      "step": 7436
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023236612702366127,
+      "loss": 0.3644,
+      "step": 7437
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002323536737235367,
+      "loss": 0.3833,
+      "step": 7438
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002323412204234122,
+      "loss": 0.2356,
+      "step": 7439
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023232876712328765,
+      "loss": 0.3425,
+      "step": 7440
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002323163138231631,
+      "loss": 0.4012,
+      "step": 7441
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002323038605230386,
+      "loss": 0.2434,
+      "step": 7442
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023229140722291403,
+      "loss": 0.4662,
+      "step": 7443
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023227895392278952,
+      "loss": 0.2858,
+      "step": 7444
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023226650062266497,
+      "loss": 0.2468,
+      "step": 7445
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023225404732254046,
+      "loss": 0.2988,
+      "step": 7446
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00023224159402241592,
+      "loss": 0.1793,
+      "step": 7447
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002322291407222914,
+      "loss": 0.2247,
+      "step": 7448
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023221668742216684,
+      "loss": 0.2269,
+      "step": 7449
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023220423412204232,
+      "loss": 0.3952,
+      "step": 7450
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023219178082191778,
+      "loss": 0.2103,
+      "step": 7451
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023217932752179327,
+      "loss": 0.3135,
+      "step": 7452
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023216687422166873,
+      "loss": 0.1756,
+      "step": 7453
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023215442092154422,
+      "loss": 0.2279,
+      "step": 7454
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023214196762141965,
+      "loss": 0.4098,
+      "step": 7455
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0002321295143212951,
+      "loss": 0.3024,
+      "step": 7456
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0002321170610211706,
+      "loss": 0.1976,
+      "step": 7457
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023210460772104605,
+      "loss": 0.2574,
+      "step": 7458
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023209215442092154,
+      "loss": 0.2319,
+      "step": 7459
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023207970112079697,
+      "loss": 0.2363,
+      "step": 7460
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023206724782067246,
+      "loss": 0.1548,
+      "step": 7461
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023205479452054792,
+      "loss": 0.4053,
+      "step": 7462
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0002320423412204234,
+      "loss": 0.2177,
+      "step": 7463
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023202988792029886,
+      "loss": 0.3228,
+      "step": 7464
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023201743462017435,
+      "loss": 0.2318,
+      "step": 7465
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023200498132004978,
+      "loss": 0.2625,
+      "step": 7466
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023199252801992527,
+      "loss": 0.2642,
+      "step": 7467
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023198007471980073,
+      "loss": 0.4919,
+      "step": 7468
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0002319676214196762,
+      "loss": 0.2016,
+      "step": 7469
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023195516811955167,
+      "loss": 0.2145,
+      "step": 7470
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023194271481942716,
+      "loss": 0.2384,
+      "step": 7471
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0002319302615193026,
+      "loss": 0.2668,
+      "step": 7472
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023191780821917805,
+      "loss": 0.1777,
+      "step": 7473
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00023190535491905353,
+      "loss": 0.3664,
+      "step": 7474
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.000231892901618929,
+      "loss": 0.393,
+      "step": 7475
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023188044831880448,
+      "loss": 0.2244,
+      "step": 7476
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002318679950186799,
+      "loss": 0.1856,
+      "step": 7477
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002318555417185554,
+      "loss": 0.182,
+      "step": 7478
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023184308841843086,
+      "loss": 0.1347,
+      "step": 7479
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023183063511830634,
+      "loss": 0.3761,
+      "step": 7480
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002318181818181818,
+      "loss": 0.3416,
+      "step": 7481
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002318057285180573,
+      "loss": 0.2793,
+      "step": 7482
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023179327521793272,
+      "loss": 0.3019,
+      "step": 7483
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002317808219178082,
+      "loss": 0.2757,
+      "step": 7484
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023176836861768367,
+      "loss": 0.2268,
+      "step": 7485
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023175591531755915,
+      "loss": 0.2228,
+      "step": 7486
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002317434620174346,
+      "loss": 0.5117,
+      "step": 7487
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023173100871731004,
+      "loss": 0.2374,
+      "step": 7488
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023171855541718553,
+      "loss": 0.2316,
+      "step": 7489
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.000231706102117061,
+      "loss": 0.2194,
+      "step": 7490
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023169364881693648,
+      "loss": 0.2634,
+      "step": 7491
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023168119551681193,
+      "loss": 0.1912,
+      "step": 7492
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023166874221668742,
+      "loss": 0.4553,
+      "step": 7493
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023165628891656285,
+      "loss": 0.2772,
+      "step": 7494
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023164383561643834,
+      "loss": 0.2225,
+      "step": 7495
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002316313823163138,
+      "loss": 0.3077,
+      "step": 7496
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023161892901618929,
+      "loss": 0.3299,
+      "step": 7497
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023160647571606474,
+      "loss": 0.217,
+      "step": 7498
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002315940224159402,
+      "loss": 0.2716,
+      "step": 7499
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00023158156911581566,
+      "loss": 0.3673,
+      "step": 7500
+    },
+    {
+      "epoch": 2.87,
+      "eval_cer": 0.10857053670718146,
+      "eval_loss": 0.379964143037796,
+      "eval_runtime": 47.4087,
+      "eval_samples_per_second": 36.976,
+      "eval_steps_per_second": 4.64,
+      "eval_wer": 0.18425578132778206,
+      "step": 7500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023156911581569115,
+      "loss": 0.2852,
+      "step": 7501
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0002315566625155666,
+      "loss": 0.2079,
+      "step": 7502
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023154420921544207,
+      "loss": 0.302,
+      "step": 7503
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023153175591531755,
+      "loss": 0.2313,
+      "step": 7504
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023151930261519299,
+      "loss": 0.2932,
+      "step": 7505
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023150684931506847,
+      "loss": 0.4399,
+      "step": 7506
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023149439601494393,
+      "loss": 0.4397,
+      "step": 7507
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023148194271481942,
+      "loss": 0.285,
+      "step": 7508
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023146948941469488,
+      "loss": 0.2013,
+      "step": 7509
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023145703611457034,
+      "loss": 0.2453,
+      "step": 7510
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0002314445828144458,
+      "loss": 0.296,
+      "step": 7511
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023143212951432128,
+      "loss": 0.388,
+      "step": 7512
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023141967621419674,
+      "loss": 0.2074,
+      "step": 7513
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023140722291407223,
+      "loss": 0.2509,
+      "step": 7514
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023139476961394769,
+      "loss": 0.2737,
+      "step": 7515
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023138231631382314,
+      "loss": 0.1958,
+      "step": 7516
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0002313698630136986,
+      "loss": 0.397,
+      "step": 7517
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0002313574097135741,
+      "loss": 0.4057,
+      "step": 7518
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023134495641344955,
+      "loss": 0.2999,
+      "step": 7519
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.000231332503113325,
+      "loss": 0.344,
+      "step": 7520
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023132004981320047,
+      "loss": 0.2862,
+      "step": 7521
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023130759651307593,
+      "loss": 0.2092,
+      "step": 7522
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0002312951432129514,
+      "loss": 0.2852,
+      "step": 7523
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023128268991282687,
+      "loss": 0.2526,
+      "step": 7524
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023127023661270236,
+      "loss": 0.399,
+      "step": 7525
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00023125778331257782,
+      "loss": 0.2565,
+      "step": 7526
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023124533001245328,
+      "loss": 0.1932,
+      "step": 7527
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023123287671232874,
+      "loss": 0.3065,
+      "step": 7528
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023122042341220422,
+      "loss": 0.1945,
+      "step": 7529
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023120797011207968,
+      "loss": 0.335,
+      "step": 7530
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023119551681195517,
+      "loss": 0.352,
+      "step": 7531
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0002311830635118306,
+      "loss": 0.2343,
+      "step": 7532
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023117061021170609,
+      "loss": 0.242,
+      "step": 7533
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023115815691158155,
+      "loss": 0.2098,
+      "step": 7534
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000231145703611457,
+      "loss": 0.1585,
+      "step": 7535
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0002311332503113325,
+      "loss": 0.2167,
+      "step": 7536
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023112079701120795,
+      "loss": 0.3597,
+      "step": 7537
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0002311083437110834,
+      "loss": 0.352,
+      "step": 7538
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023109589041095887,
+      "loss": 0.3458,
+      "step": 7539
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023108343711083435,
+      "loss": 0.2395,
+      "step": 7540
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023107098381070981,
+      "loss": 0.2468,
+      "step": 7541
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0002310585305105853,
+      "loss": 0.2129,
+      "step": 7542
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023104607721046073,
+      "loss": 0.4495,
+      "step": 7543
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023103362391033622,
+      "loss": 0.2863,
+      "step": 7544
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023102117061021168,
+      "loss": 0.1976,
+      "step": 7545
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023100871731008716,
+      "loss": 0.2753,
+      "step": 7546
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023099626400996262,
+      "loss": 0.1865,
+      "step": 7547
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0002309838107098381,
+      "loss": 0.2128,
+      "step": 7548
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023097135740971354,
+      "loss": 0.3593,
+      "step": 7549
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000230958904109589,
+      "loss": 0.4485,
+      "step": 7550
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0002309464508094645,
+      "loss": 0.3956,
+      "step": 7551
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023093399750933995,
+      "loss": 0.2538,
+      "step": 7552
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00023092154420921543,
+      "loss": 0.2481,
+      "step": 7553
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023090909090909086,
+      "loss": 0.2075,
+      "step": 7554
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023089663760896635,
+      "loss": 0.2313,
+      "step": 7555
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002308841843088418,
+      "loss": 0.3834,
+      "step": 7556
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002308717310087173,
+      "loss": 0.2197,
+      "step": 7557
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023085927770859276,
+      "loss": 0.299,
+      "step": 7558
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023084682440846824,
+      "loss": 0.3287,
+      "step": 7559
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023083437110834367,
+      "loss": 0.1682,
+      "step": 7560
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023082191780821916,
+      "loss": 0.2463,
+      "step": 7561
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023080946450809462,
+      "loss": 0.3429,
+      "step": 7562
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002307970112079701,
+      "loss": 0.2189,
+      "step": 7563
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023078455790784556,
+      "loss": 0.2551,
+      "step": 7564
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023077210460772105,
+      "loss": 0.1968,
+      "step": 7565
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023075965130759648,
+      "loss": 0.2381,
+      "step": 7566
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023074719800747194,
+      "loss": 0.2488,
+      "step": 7567
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023073474470734743,
+      "loss": 0.2444,
+      "step": 7568
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002307222914072229,
+      "loss": 0.2706,
+      "step": 7569
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023070983810709837,
+      "loss": 0.2252,
+      "step": 7570
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002306973848069738,
+      "loss": 0.2915,
+      "step": 7571
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002306849315068493,
+      "loss": 0.2262,
+      "step": 7572
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023067247820672475,
+      "loss": 0.2498,
+      "step": 7573
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023066002490660024,
+      "loss": 0.3166,
+      "step": 7574
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002306475716064757,
+      "loss": 0.3578,
+      "step": 7575
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023063511830635118,
+      "loss": 0.2615,
+      "step": 7576
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023062266500622661,
+      "loss": 0.1927,
+      "step": 7577
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002306102117061021,
+      "loss": 0.2891,
+      "step": 7578
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00023059775840597756,
+      "loss": 0.1886,
+      "step": 7579
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023058530510585305,
+      "loss": 0.2401,
+      "step": 7580
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0002305728518057285,
+      "loss": 0.2839,
+      "step": 7581
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023056039850560394,
+      "loss": 0.267,
+      "step": 7582
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023054794520547942,
+      "loss": 0.1891,
+      "step": 7583
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023053549190535488,
+      "loss": 0.1955,
+      "step": 7584
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023052303860523037,
+      "loss": 0.1947,
+      "step": 7585
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023051058530510583,
+      "loss": 0.3279,
+      "step": 7586
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023049813200498131,
+      "loss": 0.3428,
+      "step": 7587
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023048567870485675,
+      "loss": 0.2768,
+      "step": 7588
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023047322540473223,
+      "loss": 0.6392,
+      "step": 7589
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0002304607721046077,
+      "loss": 0.1739,
+      "step": 7590
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023044831880448318,
+      "loss": 0.1742,
+      "step": 7591
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023043586550435864,
+      "loss": 0.3142,
+      "step": 7592
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023042341220423412,
+      "loss": 0.3115,
+      "step": 7593
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023041095890410956,
+      "loss": 0.208,
+      "step": 7594
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023039850560398504,
+      "loss": 0.266,
+      "step": 7595
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0002303860523038605,
+      "loss": 0.2121,
+      "step": 7596
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023037359900373596,
+      "loss": 0.3193,
+      "step": 7597
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023036114570361145,
+      "loss": 0.2003,
+      "step": 7598
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023034869240348688,
+      "loss": 0.2475,
+      "step": 7599
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023033623910336237,
+      "loss": 0.4832,
+      "step": 7600
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023032378580323782,
+      "loss": 0.1897,
+      "step": 7601
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0002303113325031133,
+      "loss": 0.2377,
+      "step": 7602
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023029887920298877,
+      "loss": 0.2943,
+      "step": 7603
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00023028642590286426,
+      "loss": 0.2117,
+      "step": 7604
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0002302739726027397,
+      "loss": 0.3264,
+      "step": 7605
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023026151930261517,
+      "loss": 0.3418,
+      "step": 7606
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023024906600249063,
+      "loss": 0.4615,
+      "step": 7607
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023023661270236612,
+      "loss": 0.4813,
+      "step": 7608
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023022415940224158,
+      "loss": 0.1824,
+      "step": 7609
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023021170610211707,
+      "loss": 0.2571,
+      "step": 7610
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0002301992528019925,
+      "loss": 0.3578,
+      "step": 7611
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023018679950186798,
+      "loss": 0.3953,
+      "step": 7612
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023017434620174344,
+      "loss": 0.2349,
+      "step": 7613
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0002301618929016189,
+      "loss": 0.2827,
+      "step": 7614
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0002301494396014944,
+      "loss": 0.1757,
+      "step": 7615
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023013698630136982,
+      "loss": 0.3483,
+      "step": 7616
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0002301245330012453,
+      "loss": 0.2711,
+      "step": 7617
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023011207970112077,
+      "loss": 0.3198,
+      "step": 7618
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023009962640099625,
+      "loss": 0.3073,
+      "step": 7619
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0002300871731008717,
+      "loss": 0.2726,
+      "step": 7620
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0002300747198007472,
+      "loss": 0.1877,
+      "step": 7621
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023006226650062263,
+      "loss": 0.296,
+      "step": 7622
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023004981320049812,
+      "loss": 0.2794,
+      "step": 7623
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023003735990037358,
+      "loss": 0.3252,
+      "step": 7624
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023002490660024906,
+      "loss": 0.5419,
+      "step": 7625
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023001245330012452,
+      "loss": 0.4287,
+      "step": 7626
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00023,
+      "loss": 0.2823,
+      "step": 7627
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00022998754669987544,
+      "loss": 0.1942,
+      "step": 7628
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0002299750933997509,
+      "loss": 0.257,
+      "step": 7629
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00022996264009962638,
+      "loss": 0.2799,
+      "step": 7630
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00022995018679950184,
+      "loss": 0.5339,
+      "step": 7631
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022993773349937733,
+      "loss": 0.2773,
+      "step": 7632
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022992528019925276,
+      "loss": 0.1965,
+      "step": 7633
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022991282689912825,
+      "loss": 0.2388,
+      "step": 7634
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0002299003735990037,
+      "loss": 0.2644,
+      "step": 7635
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0002298879202988792,
+      "loss": 0.1892,
+      "step": 7636
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022987546699875465,
+      "loss": 0.3785,
+      "step": 7637
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022986301369863014,
+      "loss": 0.2697,
+      "step": 7638
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022985056039850557,
+      "loss": 0.1805,
+      "step": 7639
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022983810709838106,
+      "loss": 0.242,
+      "step": 7640
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022982565379825652,
+      "loss": 0.212,
+      "step": 7641
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000229813200498132,
+      "loss": 0.3251,
+      "step": 7642
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022980074719800746,
+      "loss": 0.5006,
+      "step": 7643
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0002297882938978829,
+      "loss": 0.2754,
+      "step": 7644
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022977584059775838,
+      "loss": 0.2181,
+      "step": 7645
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022976338729763384,
+      "loss": 0.1539,
+      "step": 7646
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022975093399750933,
+      "loss": 0.1795,
+      "step": 7647
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022973848069738479,
+      "loss": 0.3089,
+      "step": 7648
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022972602739726027,
+      "loss": 0.2747,
+      "step": 7649
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0002297135740971357,
+      "loss": 0.415,
+      "step": 7650
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0002297011207970112,
+      "loss": 0.1926,
+      "step": 7651
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022968866749688665,
+      "loss": 0.2265,
+      "step": 7652
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022967621419676214,
+      "loss": 0.24,
+      "step": 7653
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0002296637608966376,
+      "loss": 0.2481,
+      "step": 7654
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00022965130759651308,
+      "loss": 0.486,
+      "step": 7655
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0002296388542963885,
+      "loss": 0.3526,
+      "step": 7656
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000229626400996264,
+      "loss": 0.2547,
+      "step": 7657
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022961394769613946,
+      "loss": 0.2705,
+      "step": 7658
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022960149439601494,
+      "loss": 0.2513,
+      "step": 7659
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0002295890410958904,
+      "loss": 0.1699,
+      "step": 7660
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022957658779576584,
+      "loss": 0.3593,
+      "step": 7661
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022956413449564132,
+      "loss": 0.4048,
+      "step": 7662
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022955168119551678,
+      "loss": 0.2735,
+      "step": 7663
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022953922789539227,
+      "loss": 0.1812,
+      "step": 7664
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022952677459526773,
+      "loss": 0.1636,
+      "step": 7665
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0002295143212951432,
+      "loss": 0.2652,
+      "step": 7666
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022950186799501864,
+      "loss": 0.1905,
+      "step": 7667
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022948941469489413,
+      "loss": 0.6545,
+      "step": 7668
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0002294769613947696,
+      "loss": 0.2842,
+      "step": 7669
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022946450809464508,
+      "loss": 0.2236,
+      "step": 7670
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022945205479452054,
+      "loss": 0.2324,
+      "step": 7671
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.000229439601494396,
+      "loss": 0.2048,
+      "step": 7672
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022942714819427145,
+      "loss": 0.3946,
+      "step": 7673
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022941469489414694,
+      "loss": 0.3263,
+      "step": 7674
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0002294022415940224,
+      "loss": 0.5162,
+      "step": 7675
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022938978829389786,
+      "loss": 0.2997,
+      "step": 7676
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022937733499377334,
+      "loss": 0.2834,
+      "step": 7677
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022936488169364878,
+      "loss": 0.17,
+      "step": 7678
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022935242839352426,
+      "loss": 0.1817,
+      "step": 7679
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022933997509339972,
+      "loss": 0.2257,
+      "step": 7680
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0002293275217932752,
+      "loss": 0.3101,
+      "step": 7681
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022931506849315067,
+      "loss": 0.4126,
+      "step": 7682
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00022930261519302613,
+      "loss": 0.3268,
+      "step": 7683
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022929016189290159,
+      "loss": 0.263,
+      "step": 7684
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022927770859277707,
+      "loss": 0.2211,
+      "step": 7685
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022926525529265253,
+      "loss": 0.2302,
+      "step": 7686
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022925280199252802,
+      "loss": 0.9348,
+      "step": 7687
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022924034869240348,
+      "loss": 0.3274,
+      "step": 7688
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022922789539227894,
+      "loss": 0.2778,
+      "step": 7689
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002292154420921544,
+      "loss": 0.2567,
+      "step": 7690
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022920298879202985,
+      "loss": 0.1843,
+      "step": 7691
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022919053549190534,
+      "loss": 0.2379,
+      "step": 7692
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002291780821917808,
+      "loss": 0.348,
+      "step": 7693
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022916562889165626,
+      "loss": 0.2861,
+      "step": 7694
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022915317559153172,
+      "loss": 0.2148,
+      "step": 7695
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002291407222914072,
+      "loss": 0.2378,
+      "step": 7696
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022912826899128266,
+      "loss": 0.2664,
+      "step": 7697
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022911581569115815,
+      "loss": 0.1705,
+      "step": 7698
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002291033623910336,
+      "loss": 0.2229,
+      "step": 7699
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022909090909090907,
+      "loss": 0.3839,
+      "step": 7700
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022907845579078453,
+      "loss": 0.2229,
+      "step": 7701
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022906600249066001,
+      "loss": 0.4089,
+      "step": 7702
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022905354919053547,
+      "loss": 0.2985,
+      "step": 7703
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022904109589041096,
+      "loss": 0.1923,
+      "step": 7704
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002290286425902864,
+      "loss": 0.2285,
+      "step": 7705
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022901618929016188,
+      "loss": 0.4922,
+      "step": 7706
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022900373599003734,
+      "loss": 0.2632,
+      "step": 7707
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002289912826899128,
+      "loss": 0.2947,
+      "step": 7708
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00022897882938978828,
+      "loss": 0.2503,
+      "step": 7709
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022896637608966371,
+      "loss": 0.185,
+      "step": 7710
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002289539227895392,
+      "loss": 0.2231,
+      "step": 7711
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022894146948941466,
+      "loss": 0.281,
+      "step": 7712
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022892901618929015,
+      "loss": 0.244,
+      "step": 7713
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002289165628891656,
+      "loss": 0.2339,
+      "step": 7714
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002289041095890411,
+      "loss": 0.2301,
+      "step": 7715
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022889165628891652,
+      "loss": 0.2792,
+      "step": 7716
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.000228879202988792,
+      "loss": 0.2728,
+      "step": 7717
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022886674968866747,
+      "loss": 0.3692,
+      "step": 7718
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022885429638854296,
+      "loss": 0.2522,
+      "step": 7719
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022884184308841841,
+      "loss": 0.1704,
+      "step": 7720
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002288293897882939,
+      "loss": 0.2184,
+      "step": 7721
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022881693648816933,
+      "loss": 0.1919,
+      "step": 7722
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002288044831880448,
+      "loss": 0.277,
+      "step": 7723
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022879202988792028,
+      "loss": 0.4129,
+      "step": 7724
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022877957658779574,
+      "loss": 0.666,
+      "step": 7725
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022876712328767122,
+      "loss": 0.3013,
+      "step": 7726
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022875466998754666,
+      "loss": 0.1833,
+      "step": 7727
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022874221668742214,
+      "loss": 0.2699,
+      "step": 7728
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002287297633872976,
+      "loss": 0.2329,
+      "step": 7729
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002287173100871731,
+      "loss": 0.286,
+      "step": 7730
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022870485678704855,
+      "loss": 0.3462,
+      "step": 7731
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022869240348692403,
+      "loss": 0.2681,
+      "step": 7732
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022867995018679946,
+      "loss": 0.2065,
+      "step": 7733
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00022866749688667495,
+      "loss": 0.2119,
+      "step": 7734
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002286550435865504,
+      "loss": 0.2267,
+      "step": 7735
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0002286425902864259,
+      "loss": 0.2894,
+      "step": 7736
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022863013698630136,
+      "loss": 0.3737,
+      "step": 7737
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0002286176836861768,
+      "loss": 0.2234,
+      "step": 7738
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022860523038605227,
+      "loss": 0.2797,
+      "step": 7739
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022859277708592773,
+      "loss": 0.5187,
+      "step": 7740
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022858032378580322,
+      "loss": 0.1807,
+      "step": 7741
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022856787048567868,
+      "loss": 0.2396,
+      "step": 7742
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022855541718555416,
+      "loss": 0.4864,
+      "step": 7743
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0002285429638854296,
+      "loss": 0.2233,
+      "step": 7744
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022853051058530508,
+      "loss": 0.2795,
+      "step": 7745
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022851805728518054,
+      "loss": 0.1898,
+      "step": 7746
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022850560398505603,
+      "loss": 0.2217,
+      "step": 7747
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0002284931506849315,
+      "loss": 0.2751,
+      "step": 7748
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022848069738480697,
+      "loss": 0.2958,
+      "step": 7749
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0002284682440846824,
+      "loss": 0.4103,
+      "step": 7750
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0002284557907845579,
+      "loss": 0.2864,
+      "step": 7751
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022844333748443335,
+      "loss": 0.1744,
+      "step": 7752
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022843088418430884,
+      "loss": 0.1725,
+      "step": 7753
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0002284184308841843,
+      "loss": 0.3794,
+      "step": 7754
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022840597758405973,
+      "loss": 0.359,
+      "step": 7755
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022839352428393522,
+      "loss": 0.5753,
+      "step": 7756
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022838107098381067,
+      "loss": 0.2088,
+      "step": 7757
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022836861768368616,
+      "loss": 0.2131,
+      "step": 7758
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022835616438356162,
+      "loss": 0.1954,
+      "step": 7759
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0002283437110834371,
+      "loss": 0.2253,
+      "step": 7760
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00022833125778331254,
+      "loss": 0.2216,
+      "step": 7761
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022831880448318802,
+      "loss": 0.4925,
+      "step": 7762
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022830635118306348,
+      "loss": 0.3294,
+      "step": 7763
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022829389788293897,
+      "loss": 0.362,
+      "step": 7764
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022828144458281443,
+      "loss": 0.3472,
+      "step": 7765
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022826899128268992,
+      "loss": 0.2354,
+      "step": 7766
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022825653798256535,
+      "loss": 0.3234,
+      "step": 7767
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022824408468244083,
+      "loss": 0.3991,
+      "step": 7768
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0002282316313823163,
+      "loss": 0.2762,
+      "step": 7769
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022821917808219175,
+      "loss": 0.2266,
+      "step": 7770
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022820672478206724,
+      "loss": 0.2118,
+      "step": 7771
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022819427148194267,
+      "loss": 0.2271,
+      "step": 7772
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022818181818181816,
+      "loss": 0.2285,
+      "step": 7773
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022816936488169362,
+      "loss": 0.3373,
+      "step": 7774
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0002281569115815691,
+      "loss": 0.4507,
+      "step": 7775
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022814445828144456,
+      "loss": 0.204,
+      "step": 7776
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022813200498132005,
+      "loss": 0.1867,
+      "step": 7777
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022811955168119548,
+      "loss": 0.1555,
+      "step": 7778
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022810709838107097,
+      "loss": 0.3484,
+      "step": 7779
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022809464508094643,
+      "loss": 0.2742,
+      "step": 7780
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0002280821917808219,
+      "loss": 0.3078,
+      "step": 7781
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022806973848069737,
+      "loss": 0.2291,
+      "step": 7782
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022805728518057286,
+      "loss": 0.3688,
+      "step": 7783
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0002280448318804483,
+      "loss": 0.2806,
+      "step": 7784
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022803237858032378,
+      "loss": 0.2903,
+      "step": 7785
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00022801992528019923,
+      "loss": 0.209,
+      "step": 7786
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0002280074719800747,
+      "loss": 0.4472,
+      "step": 7787
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022799501867995018,
+      "loss": 0.2273,
+      "step": 7788
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0002279825653798256,
+      "loss": 0.2005,
+      "step": 7789
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0002279701120797011,
+      "loss": 0.2198,
+      "step": 7790
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022795765877957656,
+      "loss": 0.1537,
+      "step": 7791
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022794520547945204,
+      "loss": 0.214,
+      "step": 7792
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0002279327521793275,
+      "loss": 0.5226,
+      "step": 7793
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.000227920298879203,
+      "loss": 0.2887,
+      "step": 7794
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022790784557907842,
+      "loss": 0.1576,
+      "step": 7795
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0002278953922789539,
+      "loss": 0.1848,
+      "step": 7796
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022788293897882937,
+      "loss": 0.3816,
+      "step": 7797
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022787048567870485,
+      "loss": 0.2609,
+      "step": 7798
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0002278580323785803,
+      "loss": 0.2427,
+      "step": 7799
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0002278455790784558,
+      "loss": 0.38,
+      "step": 7800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022783312577833123,
+      "loss": 0.2431,
+      "step": 7801
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0002278206724782067,
+      "loss": 0.2247,
+      "step": 7802
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022780821917808218,
+      "loss": 0.2291,
+      "step": 7803
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022779576587795764,
+      "loss": 0.4346,
+      "step": 7804
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022778331257783312,
+      "loss": 0.3562,
+      "step": 7805
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022777085927770855,
+      "loss": 0.3035,
+      "step": 7806
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022775840597758404,
+      "loss": 0.269,
+      "step": 7807
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0002277459526774595,
+      "loss": 0.3084,
+      "step": 7808
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022773349937733499,
+      "loss": 0.1803,
+      "step": 7809
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022772104607721044,
+      "loss": 0.2028,
+      "step": 7810
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022770859277708593,
+      "loss": 0.2284,
+      "step": 7811
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022769613947696136,
+      "loss": 0.4419,
+      "step": 7812
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00022768368617683685,
+      "loss": 0.2278,
+      "step": 7813
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002276712328767123,
+      "loss": 0.2446,
+      "step": 7814
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002276587795765878,
+      "loss": 0.1989,
+      "step": 7815
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022764632627646325,
+      "loss": 0.2652,
+      "step": 7816
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022763387297633869,
+      "loss": 0.2072,
+      "step": 7817
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022762141967621417,
+      "loss": 0.4486,
+      "step": 7818
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022760896637608963,
+      "loss": 0.2016,
+      "step": 7819
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022759651307596512,
+      "loss": 0.249,
+      "step": 7820
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022758405977584058,
+      "loss": 0.2216,
+      "step": 7821
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022757160647571606,
+      "loss": 0.3258,
+      "step": 7822
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002275591531755915,
+      "loss": 0.298,
+      "step": 7823
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022754669987546698,
+      "loss": 0.2927,
+      "step": 7824
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022753424657534244,
+      "loss": 0.3707,
+      "step": 7825
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022752179327521793,
+      "loss": 0.2164,
+      "step": 7826
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022750933997509339,
+      "loss": 0.2573,
+      "step": 7827
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022749688667496887,
+      "loss": 0.1995,
+      "step": 7828
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002274844333748443,
+      "loss": 0.2253,
+      "step": 7829
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002274719800747198,
+      "loss": 0.2608,
+      "step": 7830
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022745952677459525,
+      "loss": 0.1974,
+      "step": 7831
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022744707347447074,
+      "loss": 0.1373,
+      "step": 7832
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002274346201743462,
+      "loss": 0.3267,
+      "step": 7833
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022742216687422163,
+      "loss": 0.2706,
+      "step": 7834
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002274097135740971,
+      "loss": 0.2206,
+      "step": 7835
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022739726027397257,
+      "loss": 0.3161,
+      "step": 7836
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022738480697384806,
+      "loss": 0.2086,
+      "step": 7837
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022737235367372352,
+      "loss": 0.1884,
+      "step": 7838
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000227359900373599,
+      "loss": 0.2501,
+      "step": 7839
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00022734744707347444,
+      "loss": 0.3897,
+      "step": 7840
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022733499377334992,
+      "loss": 0.2586,
+      "step": 7841
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022732254047322538,
+      "loss": 0.2757,
+      "step": 7842
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022731008717310087,
+      "loss": 0.1387,
+      "step": 7843
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022729763387297633,
+      "loss": 0.1938,
+      "step": 7844
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022728518057285179,
+      "loss": 0.29,
+      "step": 7845
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022727272727272725,
+      "loss": 0.3165,
+      "step": 7846
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022726027397260273,
+      "loss": 0.2201,
+      "step": 7847
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002272478206724782,
+      "loss": 0.194,
+      "step": 7848
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022723536737235365,
+      "loss": 0.186,
+      "step": 7849
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022722291407222914,
+      "loss": 0.1182,
+      "step": 7850
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022721046077210457,
+      "loss": 0.1624,
+      "step": 7851
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022719800747198005,
+      "loss": 0.3515,
+      "step": 7852
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022718555417185551,
+      "loss": 0.2882,
+      "step": 7853
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.000227173100871731,
+      "loss": 0.1641,
+      "step": 7854
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022716064757160646,
+      "loss": 0.1619,
+      "step": 7855
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022714819427148192,
+      "loss": 0.1863,
+      "step": 7856
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022713574097135738,
+      "loss": 0.3417,
+      "step": 7857
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022712328767123286,
+      "loss": 0.2127,
+      "step": 7858
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022711083437110832,
+      "loss": 0.4098,
+      "step": 7859
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002270983810709838,
+      "loss": 0.2416,
+      "step": 7860
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022708592777085927,
+      "loss": 0.2833,
+      "step": 7861
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022707347447073473,
+      "loss": 0.1922,
+      "step": 7862
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002270610211706102,
+      "loss": 0.2021,
+      "step": 7863
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022704856787048565,
+      "loss": 0.2382,
+      "step": 7864
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00022703611457036113,
+      "loss": 0.4424,
+      "step": 7865
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0002270236612702366,
+      "loss": 0.2809,
+      "step": 7866
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022701120797011205,
+      "loss": 0.2986,
+      "step": 7867
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0002269987546699875,
+      "loss": 0.2381,
+      "step": 7868
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000226986301369863,
+      "loss": 0.2737,
+      "step": 7869
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022697384806973846,
+      "loss": 0.2161,
+      "step": 7870
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022696139476961394,
+      "loss": 0.237,
+      "step": 7871
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022694894146948937,
+      "loss": 0.2172,
+      "step": 7872
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022693648816936486,
+      "loss": 0.245,
+      "step": 7873
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022692403486924032,
+      "loss": 0.2003,
+      "step": 7874
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0002269115815691158,
+      "loss": 0.1663,
+      "step": 7875
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022689912826899126,
+      "loss": 0.2398,
+      "step": 7876
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022688667496886675,
+      "loss": 0.3346,
+      "step": 7877
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022687422166874218,
+      "loss": 0.2147,
+      "step": 7878
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022686176836861767,
+      "loss": 0.2247,
+      "step": 7879
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022684931506849313,
+      "loss": 0.2544,
+      "step": 7880
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0002268368617683686,
+      "loss": 0.2321,
+      "step": 7881
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022682440846824407,
+      "loss": 0.2888,
+      "step": 7882
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0002268119551681195,
+      "loss": 0.2593,
+      "step": 7883
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000226799501867995,
+      "loss": 0.2857,
+      "step": 7884
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022678704856787045,
+      "loss": 0.3215,
+      "step": 7885
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022677459526774594,
+      "loss": 0.1743,
+      "step": 7886
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0002267621419676214,
+      "loss": 0.1819,
+      "step": 7887
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022674968866749688,
+      "loss": 0.2368,
+      "step": 7888
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022673723536737231,
+      "loss": 0.217,
+      "step": 7889
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0002267247820672478,
+      "loss": 0.5441,
+      "step": 7890
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022671232876712326,
+      "loss": 0.188,
+      "step": 7891
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00022669987546699875,
+      "loss": 0.234,
+      "step": 7892
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002266874221668742,
+      "loss": 0.197,
+      "step": 7893
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002266749688667497,
+      "loss": 0.1952,
+      "step": 7894
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022666251556662512,
+      "loss": 0.1921,
+      "step": 7895
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022665006226650058,
+      "loss": 0.3191,
+      "step": 7896
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022663760896637607,
+      "loss": 0.2488,
+      "step": 7897
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022662515566625153,
+      "loss": 0.2149,
+      "step": 7898
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022661270236612702,
+      "loss": 0.1988,
+      "step": 7899
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022660024906600245,
+      "loss": 0.2317,
+      "step": 7900
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022658779576587793,
+      "loss": 0.1993,
+      "step": 7901
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002265753424657534,
+      "loss": 0.3392,
+      "step": 7902
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022656288916562888,
+      "loss": 0.2235,
+      "step": 7903
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022655043586550434,
+      "loss": 0.2052,
+      "step": 7904
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022653798256537982,
+      "loss": 0.1917,
+      "step": 7905
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022652552926525526,
+      "loss": 0.1389,
+      "step": 7906
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022651307596513074,
+      "loss": 0.1999,
+      "step": 7907
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002265006226650062,
+      "loss": 0.375,
+      "step": 7908
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002264881693648817,
+      "loss": 0.2977,
+      "step": 7909
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022647571606475715,
+      "loss": 0.2772,
+      "step": 7910
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022646326276463258,
+      "loss": 0.265,
+      "step": 7911
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022645080946450807,
+      "loss": 0.2047,
+      "step": 7912
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022643835616438352,
+      "loss": 0.1764,
+      "step": 7913
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000226425902864259,
+      "loss": 0.3011,
+      "step": 7914
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022641344956413447,
+      "loss": 0.5133,
+      "step": 7915
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022640099626400996,
+      "loss": 0.2479,
+      "step": 7916
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0002263885429638854,
+      "loss": 0.2719,
+      "step": 7917
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00022637608966376087,
+      "loss": 0.2304,
+      "step": 7918
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022636363636363633,
+      "loss": 0.2801,
+      "step": 7919
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022635118306351182,
+      "loss": 0.219,
+      "step": 7920
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022633872976338728,
+      "loss": 0.2501,
+      "step": 7921
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022632627646326277,
+      "loss": 0.324,
+      "step": 7922
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0002263138231631382,
+      "loss": 0.2442,
+      "step": 7923
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022630136986301368,
+      "loss": 0.182,
+      "step": 7924
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022628891656288914,
+      "loss": 0.17,
+      "step": 7925
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022627646326276463,
+      "loss": 0.1629,
+      "step": 7926
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0002262640099626401,
+      "loss": 0.3069,
+      "step": 7927
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022625155666251552,
+      "loss": 0.3084,
+      "step": 7928
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.000226239103362391,
+      "loss": 0.2107,
+      "step": 7929
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022622665006226647,
+      "loss": 0.262,
+      "step": 7930
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022621419676214195,
+      "loss": 0.1751,
+      "step": 7931
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0002262017434620174,
+      "loss": 0.2083,
+      "step": 7932
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0002261892901618929,
+      "loss": 0.2366,
+      "step": 7933
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022617683686176833,
+      "loss": 0.246,
+      "step": 7934
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022616438356164382,
+      "loss": 0.1866,
+      "step": 7935
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022615193026151928,
+      "loss": 0.2948,
+      "step": 7936
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022613947696139476,
+      "loss": 0.2493,
+      "step": 7937
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022612702366127022,
+      "loss": 0.4234,
+      "step": 7938
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0002261145703611457,
+      "loss": 0.1962,
+      "step": 7939
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022610211706102114,
+      "loss": 0.2734,
+      "step": 7940
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022608966376089663,
+      "loss": 0.2166,
+      "step": 7941
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022607721046077208,
+      "loss": 0.2856,
+      "step": 7942
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022606475716064754,
+      "loss": 0.2192,
+      "step": 7943
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00022605230386052303,
+      "loss": 0.2898,
+      "step": 7944
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022603985056039846,
+      "loss": 0.2185,
+      "step": 7945
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022602739726027395,
+      "loss": 0.334,
+      "step": 7946
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002260149439601494,
+      "loss": 0.279,
+      "step": 7947
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002260024906600249,
+      "loss": 0.1873,
+      "step": 7948
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022599003735990035,
+      "loss": 0.1787,
+      "step": 7949
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022597758405977584,
+      "loss": 0.2604,
+      "step": 7950
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022596513075965127,
+      "loss": 0.3434,
+      "step": 7951
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022595267745952676,
+      "loss": 0.2842,
+      "step": 7952
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022594022415940222,
+      "loss": 0.2636,
+      "step": 7953
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002259277708592777,
+      "loss": 0.3011,
+      "step": 7954
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022591531755915316,
+      "loss": 0.2815,
+      "step": 7955
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022590286425902865,
+      "loss": 0.1488,
+      "step": 7956
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022589041095890408,
+      "loss": 0.2838,
+      "step": 7957
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022587795765877954,
+      "loss": 0.3274,
+      "step": 7958
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022586550435865503,
+      "loss": 0.3125,
+      "step": 7959
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022585305105853049,
+      "loss": 0.2184,
+      "step": 7960
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022584059775840597,
+      "loss": 0.2035,
+      "step": 7961
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002258281444582814,
+      "loss": 0.1556,
+      "step": 7962
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002258156911581569,
+      "loss": 0.2549,
+      "step": 7963
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022580323785803235,
+      "loss": 0.1763,
+      "step": 7964
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022579078455790784,
+      "loss": 0.3307,
+      "step": 7965
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002257783312577833,
+      "loss": 0.2271,
+      "step": 7966
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022576587795765878,
+      "loss": 0.2041,
+      "step": 7967
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002257534246575342,
+      "loss": 0.1771,
+      "step": 7968
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0002257409713574097,
+      "loss": 0.2425,
+      "step": 7969
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00022572851805728516,
+      "loss": 0.2829,
+      "step": 7970
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022571606475716064,
+      "loss": 0.392,
+      "step": 7971
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0002257036114570361,
+      "loss": 0.2049,
+      "step": 7972
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0002256911581569116,
+      "loss": 0.3048,
+      "step": 7973
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022567870485678702,
+      "loss": 0.1775,
+      "step": 7974
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022566625155666248,
+      "loss": 0.1454,
+      "step": 7975
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022565379825653797,
+      "loss": 0.198,
+      "step": 7976
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022564134495641343,
+      "loss": 0.3823,
+      "step": 7977
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0002256288916562889,
+      "loss": 0.2623,
+      "step": 7978
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022561643835616434,
+      "loss": 0.271,
+      "step": 7979
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022560398505603983,
+      "loss": 0.1753,
+      "step": 7980
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0002255915317559153,
+      "loss": 0.3172,
+      "step": 7981
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022557907845579078,
+      "loss": 0.2126,
+      "step": 7982
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022556662515566624,
+      "loss": 0.2742,
+      "step": 7983
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022555417185554172,
+      "loss": 0.4753,
+      "step": 7984
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022554171855541715,
+      "loss": 0.1816,
+      "step": 7985
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022552926525529264,
+      "loss": 0.1772,
+      "step": 7986
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0002255168119551681,
+      "loss": 0.2017,
+      "step": 7987
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022550435865504359,
+      "loss": 0.3293,
+      "step": 7988
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022549190535491904,
+      "loss": 0.1878,
+      "step": 7989
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022547945205479448,
+      "loss": 0.3072,
+      "step": 7990
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022546699875466996,
+      "loss": 0.4067,
+      "step": 7991
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022545454545454542,
+      "loss": 0.2588,
+      "step": 7992
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0002254420921544209,
+      "loss": 0.2279,
+      "step": 7993
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022542963885429637,
+      "loss": 0.1963,
+      "step": 7994
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022541718555417185,
+      "loss": 0.2125,
+      "step": 7995
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00022540473225404729,
+      "loss": 0.3883,
+      "step": 7996
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022539227895392277,
+      "loss": 0.3379,
+      "step": 7997
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022537982565379823,
+      "loss": 0.1858,
+      "step": 7998
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022536737235367372,
+      "loss": 0.3349,
+      "step": 7999
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022535491905354918,
+      "loss": 0.3422,
+      "step": 8000
+    },
+    {
+      "epoch": 3.07,
+      "eval_cer": 0.10924556645921532,
+      "eval_loss": 0.3932405710220337,
+      "eval_runtime": 47.7352,
+      "eval_samples_per_second": 36.723,
+      "eval_steps_per_second": 4.609,
+      "eval_wer": 0.18296601348599356,
+      "step": 8000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022534246575342466,
+      "loss": 0.1787,
+      "step": 8001
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0002253300124533001,
+      "loss": 0.2644,
+      "step": 8002
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022531755915317558,
+      "loss": 0.3181,
+      "step": 8003
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022530510585305104,
+      "loss": 0.2343,
+      "step": 8004
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0002252926525529265,
+      "loss": 0.1783,
+      "step": 8005
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022528019925280199,
+      "loss": 0.1741,
+      "step": 8006
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022526774595267742,
+      "loss": 0.1792,
+      "step": 8007
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0002252552926525529,
+      "loss": 0.2507,
+      "step": 8008
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022524283935242836,
+      "loss": 0.3516,
+      "step": 8009
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022523038605230385,
+      "loss": 0.2396,
+      "step": 8010
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0002252179327521793,
+      "loss": 0.1666,
+      "step": 8011
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0002252054794520548,
+      "loss": 0.2123,
+      "step": 8012
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022519302615193023,
+      "loss": 0.2697,
+      "step": 8013
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022518057285180571,
+      "loss": 0.2426,
+      "step": 8014
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022516811955168117,
+      "loss": 0.3741,
+      "step": 8015
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022515566625155666,
+      "loss": 0.2061,
+      "step": 8016
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022514321295143212,
+      "loss": 0.2471,
+      "step": 8017
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022513075965130758,
+      "loss": 0.1585,
+      "step": 8018
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022511830635118304,
+      "loss": 0.205,
+      "step": 8019
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022510585305105852,
+      "loss": 0.1945,
+      "step": 8020
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022509339975093398,
+      "loss": 0.3653,
+      "step": 8021
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00022508094645080944,
+      "loss": 0.2022,
+      "step": 8022
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022506849315068493,
+      "loss": 0.1639,
+      "step": 8023
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022505603985056036,
+      "loss": 0.251,
+      "step": 8024
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022504358655043585,
+      "loss": 0.1608,
+      "step": 8025
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0002250311332503113,
+      "loss": 0.1761,
+      "step": 8026
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0002250186799501868,
+      "loss": 0.3182,
+      "step": 8027
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022500622665006225,
+      "loss": 0.1984,
+      "step": 8028
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0002249937733499377,
+      "loss": 0.1987,
+      "step": 8029
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022498132004981317,
+      "loss": 0.2071,
+      "step": 8030
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022496886674968866,
+      "loss": 0.1711,
+      "step": 8031
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022495641344956411,
+      "loss": 0.1352,
+      "step": 8032
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0002249439601494396,
+      "loss": 0.17,
+      "step": 8033
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022493150684931503,
+      "loss": 0.2328,
+      "step": 8034
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022491905354919052,
+      "loss": 0.1775,
+      "step": 8035
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022490660024906598,
+      "loss": 0.1761,
+      "step": 8036
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022489414694894144,
+      "loss": 0.1478,
+      "step": 8037
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022488169364881692,
+      "loss": 0.2727,
+      "step": 8038
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022486924034869238,
+      "loss": 0.2324,
+      "step": 8039
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022485678704856784,
+      "loss": 0.3026,
+      "step": 8040
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0002248443337484433,
+      "loss": 0.2226,
+      "step": 8041
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0002248318804483188,
+      "loss": 0.2766,
+      "step": 8042
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022481942714819425,
+      "loss": 0.186,
+      "step": 8043
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022480697384806973,
+      "loss": 0.2819,
+      "step": 8044
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022479452054794516,
+      "loss": 0.2194,
+      "step": 8045
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00022478206724782065,
+      "loss": 0.3138,
+      "step": 8046
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0002247696139476961,
+      "loss": 0.252,
+      "step": 8047
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0002247571606475716,
+      "loss": 0.2588,
+      "step": 8048
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022474470734744706,
+      "loss": 0.3652,
+      "step": 8049
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022473225404732254,
+      "loss": 0.1963,
+      "step": 8050
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022471980074719797,
+      "loss": 0.2011,
+      "step": 8051
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022470734744707346,
+      "loss": 0.3092,
+      "step": 8052
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022469489414694892,
+      "loss": 0.2368,
+      "step": 8053
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022468244084682438,
+      "loss": 0.1681,
+      "step": 8054
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022466998754669987,
+      "loss": 0.3173,
+      "step": 8055
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0002246575342465753,
+      "loss": 0.1688,
+      "step": 8056
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022464508094645078,
+      "loss": 0.1656,
+      "step": 8057
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022463262764632624,
+      "loss": 0.2176,
+      "step": 8058
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022462017434620173,
+      "loss": 0.2903,
+      "step": 8059
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0002246077210460772,
+      "loss": 0.3947,
+      "step": 8060
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022459526774595267,
+      "loss": 0.2276,
+      "step": 8061
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0002245828144458281,
+      "loss": 0.2274,
+      "step": 8062
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0002245703611457036,
+      "loss": 0.1584,
+      "step": 8063
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022455790784557905,
+      "loss": 0.2656,
+      "step": 8064
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022454545454545454,
+      "loss": 0.3094,
+      "step": 8065
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022453300124533,
+      "loss": 0.2279,
+      "step": 8066
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022452054794520548,
+      "loss": 0.1953,
+      "step": 8067
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022450809464508092,
+      "loss": 0.2094,
+      "step": 8068
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022449564134495637,
+      "loss": 0.2581,
+      "step": 8069
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022448318804483186,
+      "loss": 0.2155,
+      "step": 8070
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022447073474470732,
+      "loss": 0.4756,
+      "step": 8071
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0002244582814445828,
+      "loss": 0.2075,
+      "step": 8072
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022444582814445824,
+      "loss": 0.1981,
+      "step": 8073
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00022443337484433372,
+      "loss": 0.2484,
+      "step": 8074
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022442092154420918,
+      "loss": 0.1631,
+      "step": 8075
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022440846824408467,
+      "loss": 0.2434,
+      "step": 8076
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022439601494396013,
+      "loss": 0.3499,
+      "step": 8077
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022438356164383562,
+      "loss": 0.2051,
+      "step": 8078
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022437110834371105,
+      "loss": 0.2627,
+      "step": 8079
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022435865504358653,
+      "loss": 0.1894,
+      "step": 8080
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000224346201743462,
+      "loss": 0.2148,
+      "step": 8081
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022433374844333748,
+      "loss": 0.1891,
+      "step": 8082
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022432129514321294,
+      "loss": 0.3211,
+      "step": 8083
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022430884184308837,
+      "loss": 0.3946,
+      "step": 8084
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022429638854296386,
+      "loss": 0.2157,
+      "step": 8085
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022428393524283932,
+      "loss": 0.1758,
+      "step": 8086
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0002242714819427148,
+      "loss": 0.234,
+      "step": 8087
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022425902864259026,
+      "loss": 0.1707,
+      "step": 8088
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022424657534246575,
+      "loss": 0.3446,
+      "step": 8089
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022423412204234118,
+      "loss": 0.2842,
+      "step": 8090
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022422166874221667,
+      "loss": 0.2127,
+      "step": 8091
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022420921544209213,
+      "loss": 0.1652,
+      "step": 8092
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0002241967621419676,
+      "loss": 0.1768,
+      "step": 8093
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022418430884184307,
+      "loss": 0.2297,
+      "step": 8094
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022417185554171856,
+      "loss": 0.2884,
+      "step": 8095
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.000224159402241594,
+      "loss": 0.5171,
+      "step": 8096
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022414694894146948,
+      "loss": 0.1987,
+      "step": 8097
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022413449564134493,
+      "loss": 0.3237,
+      "step": 8098
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022412204234122042,
+      "loss": 0.1561,
+      "step": 8099
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00022410958904109588,
+      "loss": 0.2392,
+      "step": 8100
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002240971357409713,
+      "loss": 0.2309,
+      "step": 8101
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002240846824408468,
+      "loss": 0.3962,
+      "step": 8102
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022407222914072226,
+      "loss": 0.2944,
+      "step": 8103
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022405977584059774,
+      "loss": 0.1984,
+      "step": 8104
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002240473225404732,
+      "loss": 0.2264,
+      "step": 8105
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002240348692403487,
+      "loss": 0.3114,
+      "step": 8106
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022402241594022412,
+      "loss": 0.1858,
+      "step": 8107
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002240099626400996,
+      "loss": 0.2261,
+      "step": 8108
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022399750933997507,
+      "loss": 0.252,
+      "step": 8109
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022398505603985055,
+      "loss": 0.2396,
+      "step": 8110
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.000223972602739726,
+      "loss": 0.6345,
+      "step": 8111
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002239601494396015,
+      "loss": 0.4311,
+      "step": 8112
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022394769613947693,
+      "loss": 0.1834,
+      "step": 8113
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022393524283935242,
+      "loss": 0.2846,
+      "step": 8114
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022392278953922788,
+      "loss": 0.2845,
+      "step": 8115
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022391033623910334,
+      "loss": 0.3314,
+      "step": 8116
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022389788293897882,
+      "loss": 0.2053,
+      "step": 8117
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022388542963885425,
+      "loss": 0.2347,
+      "step": 8118
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022387297633872974,
+      "loss": 0.1609,
+      "step": 8119
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002238605230386052,
+      "loss": 0.2617,
+      "step": 8120
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022384806973848069,
+      "loss": 0.307,
+      "step": 8121
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022383561643835614,
+      "loss": 0.1721,
+      "step": 8122
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022382316313823163,
+      "loss": 0.2578,
+      "step": 8123
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022381070983810706,
+      "loss": 0.1801,
+      "step": 8124
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00022379825653798255,
+      "loss": 0.1442,
+      "step": 8125
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.000223785803237858,
+      "loss": 0.3005,
+      "step": 8126
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002237733499377335,
+      "loss": 0.3605,
+      "step": 8127
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022376089663760895,
+      "loss": 0.3208,
+      "step": 8128
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022374844333748444,
+      "loss": 0.2816,
+      "step": 8129
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022373599003735987,
+      "loss": 0.162,
+      "step": 8130
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022372353673723533,
+      "loss": 0.1553,
+      "step": 8131
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022371108343711082,
+      "loss": 0.2067,
+      "step": 8132
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022369863013698628,
+      "loss": 0.4182,
+      "step": 8133
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022368617683686176,
+      "loss": 0.4056,
+      "step": 8134
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0002236737235367372,
+      "loss": 0.2491,
+      "step": 8135
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022366127023661268,
+      "loss": 0.3414,
+      "step": 8136
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022364881693648814,
+      "loss": 0.288,
+      "step": 8137
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022363636363636363,
+      "loss": 0.2936,
+      "step": 8138
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022362391033623909,
+      "loss": 0.1958,
+      "step": 8139
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022361145703611457,
+      "loss": 0.3771,
+      "step": 8140
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022359900373599,
+      "loss": 0.3042,
+      "step": 8141
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0002235865504358655,
+      "loss": 0.1795,
+      "step": 8142
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022357409713574095,
+      "loss": 0.3448,
+      "step": 8143
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022356164383561644,
+      "loss": 0.3509,
+      "step": 8144
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0002235491905354919,
+      "loss": 0.1173,
+      "step": 8145
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022353673723536738,
+      "loss": 0.2736,
+      "step": 8146
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0002235242839352428,
+      "loss": 0.2482,
+      "step": 8147
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022351183063511827,
+      "loss": 0.1908,
+      "step": 8148
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022349937733499376,
+      "loss": 0.1976,
+      "step": 8149
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022348692403486922,
+      "loss": 0.2128,
+      "step": 8150
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0002234744707347447,
+      "loss": 0.1703,
+      "step": 8151
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022346201743462014,
+      "loss": 0.2747,
+      "step": 8152
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00022344956413449562,
+      "loss": 0.3856,
+      "step": 8153
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022343711083437108,
+      "loss": 0.2401,
+      "step": 8154
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022342465753424657,
+      "loss": 0.2529,
+      "step": 8155
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022341220423412203,
+      "loss": 0.216,
+      "step": 8156
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0002233997509339975,
+      "loss": 0.2322,
+      "step": 8157
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022338729763387295,
+      "loss": 0.2822,
+      "step": 8158
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022337484433374843,
+      "loss": 0.4793,
+      "step": 8159
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0002233623910336239,
+      "loss": 0.1933,
+      "step": 8160
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022334993773349938,
+      "loss": 0.2685,
+      "step": 8161
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022333748443337484,
+      "loss": 0.2016,
+      "step": 8162
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022332503113325027,
+      "loss": 0.2353,
+      "step": 8163
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022331257783312575,
+      "loss": 0.222,
+      "step": 8164
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022330012453300121,
+      "loss": 0.334,
+      "step": 8165
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0002232876712328767,
+      "loss": 0.2037,
+      "step": 8166
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022327521793275216,
+      "loss": 0.2315,
+      "step": 8167
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022326276463262765,
+      "loss": 0.1843,
+      "step": 8168
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022325031133250308,
+      "loss": 0.2071,
+      "step": 8169
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022323785803237856,
+      "loss": 0.1229,
+      "step": 8170
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022322540473225402,
+      "loss": 0.463,
+      "step": 8171
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0002232129514321295,
+      "loss": 0.2003,
+      "step": 8172
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022320049813200497,
+      "loss": 0.2479,
+      "step": 8173
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022318804483188045,
+      "loss": 0.1956,
+      "step": 8174
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0002231755915317559,
+      "loss": 0.2819,
+      "step": 8175
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022316313823163137,
+      "loss": 0.2014,
+      "step": 8176
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022315068493150683,
+      "loss": 0.4355,
+      "step": 8177
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0002231382316313823,
+      "loss": 0.2185,
+      "step": 8178
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00022312577833125778,
+      "loss": 0.1887,
+      "step": 8179
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002231133250311332,
+      "loss": 0.2132,
+      "step": 8180
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002231008717310087,
+      "loss": 0.1704,
+      "step": 8181
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022308841843088416,
+      "loss": 0.1867,
+      "step": 8182
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022307596513075964,
+      "loss": 0.2946,
+      "step": 8183
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002230635118306351,
+      "loss": 0.2979,
+      "step": 8184
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002230510585305106,
+      "loss": 0.2125,
+      "step": 8185
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022303860523038602,
+      "loss": 0.1987,
+      "step": 8186
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002230261519302615,
+      "loss": 0.1783,
+      "step": 8187
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022301369863013696,
+      "loss": 0.1911,
+      "step": 8188
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022300124533001245,
+      "loss": 0.2173,
+      "step": 8189
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002229887920298879,
+      "loss": 0.3782,
+      "step": 8190
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022297633872976337,
+      "loss": 0.2764,
+      "step": 8191
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022296388542963883,
+      "loss": 0.2573,
+      "step": 8192
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022295143212951431,
+      "loss": 0.2513,
+      "step": 8193
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022293897882938977,
+      "loss": 0.2685,
+      "step": 8194
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022292652552926523,
+      "loss": 0.2171,
+      "step": 8195
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002229140722291407,
+      "loss": 0.2871,
+      "step": 8196
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022290161892901615,
+      "loss": 0.3657,
+      "step": 8197
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022288916562889164,
+      "loss": 0.2009,
+      "step": 8198
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002228767123287671,
+      "loss": 0.1799,
+      "step": 8199
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022286425902864258,
+      "loss": 0.189,
+      "step": 8200
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022285180572851804,
+      "loss": 0.1904,
+      "step": 8201
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002228393524283935,
+      "loss": 0.3263,
+      "step": 8202
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022282689912826896,
+      "loss": 0.3113,
+      "step": 8203
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00022281444582814445,
+      "loss": 0.2339,
+      "step": 8204
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002228019925280199,
+      "loss": 0.3158,
+      "step": 8205
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0002227895392278954,
+      "loss": 0.1662,
+      "step": 8206
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022277708592777082,
+      "loss": 0.177,
+      "step": 8207
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0002227646326276463,
+      "loss": 0.3936,
+      "step": 8208
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022275217932752177,
+      "loss": 0.2411,
+      "step": 8209
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022273972602739723,
+      "loss": 0.2624,
+      "step": 8210
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022272727272727272,
+      "loss": 0.2261,
+      "step": 8211
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022271481942714817,
+      "loss": 0.1306,
+      "step": 8212
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022270236612702363,
+      "loss": 0.2253,
+      "step": 8213
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0002226899128268991,
+      "loss": 0.2305,
+      "step": 8214
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022267745952677458,
+      "loss": 0.6184,
+      "step": 8215
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022266500622665004,
+      "loss": 0.2415,
+      "step": 8216
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022265255292652552,
+      "loss": 0.2316,
+      "step": 8217
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022264009962640096,
+      "loss": 0.21,
+      "step": 8218
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022262764632627644,
+      "loss": 0.2022,
+      "step": 8219
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0002226151930261519,
+      "loss": 0.2374,
+      "step": 8220
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0002226027397260274,
+      "loss": 0.4332,
+      "step": 8221
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022259028642590285,
+      "loss": 0.2487,
+      "step": 8222
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022257783312577833,
+      "loss": 0.2367,
+      "step": 8223
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022256537982565377,
+      "loss": 0.1825,
+      "step": 8224
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022255292652552922,
+      "loss": 0.317,
+      "step": 8225
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0002225404732254047,
+      "loss": 0.1698,
+      "step": 8226
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022252801992528017,
+      "loss": 0.3238,
+      "step": 8227
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022251556662515566,
+      "loss": 0.2431,
+      "step": 8228
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0002225031133250311,
+      "loss": 0.2227,
+      "step": 8229
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022249066002490657,
+      "loss": 0.152,
+      "step": 8230
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00022247820672478203,
+      "loss": 0.1686,
+      "step": 8231
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022246575342465752,
+      "loss": 0.3094,
+      "step": 8232
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022245330012453298,
+      "loss": 0.3723,
+      "step": 8233
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022244084682440847,
+      "loss": 0.4055,
+      "step": 8234
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0002224283935242839,
+      "loss": 0.3058,
+      "step": 8235
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022241594022415938,
+      "loss": 0.2425,
+      "step": 8236
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022240348692403484,
+      "loss": 0.3056,
+      "step": 8237
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022239103362391033,
+      "loss": 0.2295,
+      "step": 8238
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0002223785803237858,
+      "loss": 0.2159,
+      "step": 8239
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022236612702366127,
+      "loss": 0.3246,
+      "step": 8240
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0002223536737235367,
+      "loss": 0.1915,
+      "step": 8241
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022234122042341217,
+      "loss": 0.1974,
+      "step": 8242
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022232876712328765,
+      "loss": 0.3081,
+      "step": 8243
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0002223163138231631,
+      "loss": 0.1871,
+      "step": 8244
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0002223038605230386,
+      "loss": 0.2202,
+      "step": 8245
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022229140722291403,
+      "loss": 0.4565,
+      "step": 8246
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022227895392278952,
+      "loss": 0.3075,
+      "step": 8247
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022226650062266498,
+      "loss": 0.2362,
+      "step": 8248
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022225404732254046,
+      "loss": 0.3541,
+      "step": 8249
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022224159402241592,
+      "loss": 0.2376,
+      "step": 8250
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0002222291407222914,
+      "loss": 0.2146,
+      "step": 8251
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022221668742216684,
+      "loss": 0.3376,
+      "step": 8252
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022220423412204233,
+      "loss": 0.3259,
+      "step": 8253
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022219178082191778,
+      "loss": 0.2044,
+      "step": 8254
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022217932752179327,
+      "loss": 0.187,
+      "step": 8255
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022216687422166873,
+      "loss": 0.2937,
+      "step": 8256
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00022215442092154416,
+      "loss": 0.2638,
+      "step": 8257
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022214196762141965,
+      "loss": 0.2785,
+      "step": 8258
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002221295143212951,
+      "loss": 0.2949,
+      "step": 8259
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002221170610211706,
+      "loss": 0.2053,
+      "step": 8260
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022210460772104605,
+      "loss": 0.2539,
+      "step": 8261
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022209215442092154,
+      "loss": 0.2385,
+      "step": 8262
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022207970112079697,
+      "loss": 0.1998,
+      "step": 8263
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022206724782067246,
+      "loss": 0.1756,
+      "step": 8264
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022205479452054792,
+      "loss": 0.337,
+      "step": 8265
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002220423412204234,
+      "loss": 0.2052,
+      "step": 8266
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022202988792029886,
+      "loss": 0.1557,
+      "step": 8267
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022201743462017435,
+      "loss": 0.1816,
+      "step": 8268
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022200498132004978,
+      "loss": 0.1365,
+      "step": 8269
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022199252801992527,
+      "loss": 0.2111,
+      "step": 8270
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022198007471980073,
+      "loss": 0.3009,
+      "step": 8271
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022196762141967619,
+      "loss": 0.2607,
+      "step": 8272
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022195516811955167,
+      "loss": 0.1747,
+      "step": 8273
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002219427148194271,
+      "loss": 0.199,
+      "step": 8274
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002219302615193026,
+      "loss": 0.1144,
+      "step": 8275
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022191780821917805,
+      "loss": 0.2432,
+      "step": 8276
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022190535491905354,
+      "loss": 0.3908,
+      "step": 8277
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.000221892901618929,
+      "loss": 0.258,
+      "step": 8278
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022188044831880448,
+      "loss": 0.2073,
+      "step": 8279
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002218679950186799,
+      "loss": 0.1835,
+      "step": 8280
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002218555417185554,
+      "loss": 0.2643,
+      "step": 8281
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022184308841843086,
+      "loss": 0.3215,
+      "step": 8282
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00022183063511830634,
+      "loss": 0.3134,
+      "step": 8283
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0002218181818181818,
+      "loss": 0.3681,
+      "step": 8284
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0002218057285180573,
+      "loss": 0.2222,
+      "step": 8285
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022179327521793272,
+      "loss": 0.2017,
+      "step": 8286
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0002217808219178082,
+      "loss": 0.1883,
+      "step": 8287
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022176836861768367,
+      "loss": 0.1976,
+      "step": 8288
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022175591531755913,
+      "loss": 0.3107,
+      "step": 8289
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0002217434620174346,
+      "loss": 0.3965,
+      "step": 8290
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022173100871731004,
+      "loss": 0.2121,
+      "step": 8291
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022171855541718553,
+      "loss": 0.2662,
+      "step": 8292
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.000221706102117061,
+      "loss": 0.3104,
+      "step": 8293
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022169364881693648,
+      "loss": 0.1782,
+      "step": 8294
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022168119551681194,
+      "loss": 0.1933,
+      "step": 8295
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022166874221668742,
+      "loss": 0.3968,
+      "step": 8296
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022165628891656285,
+      "loss": 0.2074,
+      "step": 8297
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022164383561643834,
+      "loss": 0.2224,
+      "step": 8298
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0002216313823163138,
+      "loss": 0.1949,
+      "step": 8299
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022161892901618929,
+      "loss": 0.2483,
+      "step": 8300
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022160647571606475,
+      "loss": 0.15,
+      "step": 8301
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022159402241594023,
+      "loss": 0.2892,
+      "step": 8302
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022158156911581566,
+      "loss": 0.2177,
+      "step": 8303
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022156911581569112,
+      "loss": 0.2622,
+      "step": 8304
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0002215566625155666,
+      "loss": 0.223,
+      "step": 8305
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022154420921544207,
+      "loss": 0.3617,
+      "step": 8306
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022153175591531755,
+      "loss": 0.155,
+      "step": 8307
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022151930261519299,
+      "loss": 0.1787,
+      "step": 8308
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00022150684931506847,
+      "loss": 0.3324,
+      "step": 8309
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022149439601494393,
+      "loss": 0.2991,
+      "step": 8310
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022148194271481942,
+      "loss": 0.1796,
+      "step": 8311
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022146948941469488,
+      "loss": 0.1978,
+      "step": 8312
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022145703611457036,
+      "loss": 0.2818,
+      "step": 8313
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0002214445828144458,
+      "loss": 0.2716,
+      "step": 8314
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022143212951432128,
+      "loss": 0.6577,
+      "step": 8315
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022141967621419674,
+      "loss": 0.2024,
+      "step": 8316
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022140722291407223,
+      "loss": 0.2508,
+      "step": 8317
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0002213947696139477,
+      "loss": 0.2033,
+      "step": 8318
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022138231631382317,
+      "loss": 0.2924,
+      "step": 8319
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0002213698630136986,
+      "loss": 0.1926,
+      "step": 8320
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022135740971357406,
+      "loss": 0.3798,
+      "step": 8321
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022134495641344955,
+      "loss": 0.1771,
+      "step": 8322
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.000221332503113325,
+      "loss": 0.2642,
+      "step": 8323
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0002213200498132005,
+      "loss": 0.3,
+      "step": 8324
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022130759651307593,
+      "loss": 0.2283,
+      "step": 8325
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022129514321295141,
+      "loss": 0.2774,
+      "step": 8326
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022128268991282687,
+      "loss": 0.4097,
+      "step": 8327
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022127023661270236,
+      "loss": 0.2191,
+      "step": 8328
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022125778331257782,
+      "loss": 0.325,
+      "step": 8329
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0002212453300124533,
+      "loss": 0.2715,
+      "step": 8330
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022123287671232874,
+      "loss": 0.187,
+      "step": 8331
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022122042341220422,
+      "loss": 0.2326,
+      "step": 8332
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022120797011207968,
+      "loss": 0.2897,
+      "step": 8333
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022119551681195517,
+      "loss": 0.4086,
+      "step": 8334
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00022118306351183063,
+      "loss": 0.2535,
+      "step": 8335
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022117061021170606,
+      "loss": 0.2179,
+      "step": 8336
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022115815691158155,
+      "loss": 0.2263,
+      "step": 8337
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.000221145703611457,
+      "loss": 0.2684,
+      "step": 8338
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0002211332503113325,
+      "loss": 0.4636,
+      "step": 8339
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022112079701120795,
+      "loss": 0.3278,
+      "step": 8340
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022110834371108344,
+      "loss": 0.1963,
+      "step": 8341
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022109589041095887,
+      "loss": 0.2443,
+      "step": 8342
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022108343711083436,
+      "loss": 0.1842,
+      "step": 8343
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022107098381070981,
+      "loss": 0.2436,
+      "step": 8344
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0002210585305105853,
+      "loss": 0.1789,
+      "step": 8345
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022104607721046076,
+      "loss": 0.501,
+      "step": 8346
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022103362391033625,
+      "loss": 0.1886,
+      "step": 8347
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022102117061021168,
+      "loss": 0.1624,
+      "step": 8348
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022100871731008716,
+      "loss": 0.2269,
+      "step": 8349
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022099626400996262,
+      "loss": 0.1839,
+      "step": 8350
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022098381070983808,
+      "loss": 0.2349,
+      "step": 8351
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022097135740971357,
+      "loss": 0.4161,
+      "step": 8352
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.000220958904109589,
+      "loss": 0.2141,
+      "step": 8353
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0002209464508094645,
+      "loss": 0.2593,
+      "step": 8354
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022093399750933995,
+      "loss": 0.1991,
+      "step": 8355
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022092154420921543,
+      "loss": 0.2221,
+      "step": 8356
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0002209090909090909,
+      "loss": 0.2023,
+      "step": 8357
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022089663760896635,
+      "loss": 0.3674,
+      "step": 8358
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0002208841843088418,
+      "loss": 0.2603,
+      "step": 8359
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0002208717310087173,
+      "loss": 0.2539,
+      "step": 8360
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00022085927770859276,
+      "loss": 0.2291,
+      "step": 8361
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022084682440846824,
+      "loss": 0.1639,
+      "step": 8362
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002208343711083437,
+      "loss": 0.1622,
+      "step": 8363
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022082191780821916,
+      "loss": 0.2026,
+      "step": 8364
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022080946450809462,
+      "loss": 0.2961,
+      "step": 8365
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002207970112079701,
+      "loss": 0.2047,
+      "step": 8366
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022078455790784557,
+      "loss": 0.3073,
+      "step": 8367
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022077210460772102,
+      "loss": 0.1754,
+      "step": 8368
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022075965130759648,
+      "loss": 0.1679,
+      "step": 8369
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022074719800747194,
+      "loss": 0.1862,
+      "step": 8370
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022073474470734743,
+      "loss": 0.3747,
+      "step": 8371
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002207222914072229,
+      "loss": 0.2003,
+      "step": 8372
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022070983810709837,
+      "loss": 0.168,
+      "step": 8373
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022069738480697383,
+      "loss": 0.3326,
+      "step": 8374
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002206849315068493,
+      "loss": 0.1509,
+      "step": 8375
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022067247820672475,
+      "loss": 0.1903,
+      "step": 8376
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022066002490660024,
+      "loss": 0.3567,
+      "step": 8377
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002206475716064757,
+      "loss": 0.2394,
+      "step": 8378
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022063511830635118,
+      "loss": 0.312,
+      "step": 8379
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022062266500622662,
+      "loss": 0.2821,
+      "step": 8380
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002206102117061021,
+      "loss": 0.2307,
+      "step": 8381
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022059775840597756,
+      "loss": 0.3189,
+      "step": 8382
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022058530510585302,
+      "loss": 0.2423,
+      "step": 8383
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002205728518057285,
+      "loss": 0.3576,
+      "step": 8384
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022056039850560397,
+      "loss": 0.1852,
+      "step": 8385
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022054794520547942,
+      "loss": 0.2608,
+      "step": 8386
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00022053549190535488,
+      "loss": 0.188,
+      "step": 8387
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022052303860523037,
+      "loss": 0.2455,
+      "step": 8388
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022051058530510583,
+      "loss": 0.3585,
+      "step": 8389
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022049813200498132,
+      "loss": 0.4703,
+      "step": 8390
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022048567870485675,
+      "loss": 0.35,
+      "step": 8391
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022047322540473223,
+      "loss": 0.2313,
+      "step": 8392
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0002204607721046077,
+      "loss": 0.1968,
+      "step": 8393
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022044831880448318,
+      "loss": 0.1495,
+      "step": 8394
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022043586550435864,
+      "loss": 0.2362,
+      "step": 8395
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022042341220423413,
+      "loss": 0.2427,
+      "step": 8396
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022041095890410956,
+      "loss": 0.2271,
+      "step": 8397
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022039850560398502,
+      "loss": 0.2124,
+      "step": 8398
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0002203860523038605,
+      "loss": 0.2727,
+      "step": 8399
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022037359900373596,
+      "loss": 0.2208,
+      "step": 8400
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022036114570361145,
+      "loss": 0.239,
+      "step": 8401
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022034869240348688,
+      "loss": 0.2392,
+      "step": 8402
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022033623910336237,
+      "loss": 0.2568,
+      "step": 8403
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022032378580323783,
+      "loss": 0.3724,
+      "step": 8404
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0002203113325031133,
+      "loss": 0.2126,
+      "step": 8405
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022029887920298877,
+      "loss": 0.2327,
+      "step": 8406
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022028642590286426,
+      "loss": 0.1814,
+      "step": 8407
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0002202739726027397,
+      "loss": 0.3595,
+      "step": 8408
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022026151930261518,
+      "loss": 0.3016,
+      "step": 8409
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022024906600249063,
+      "loss": 0.2279,
+      "step": 8410
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022023661270236612,
+      "loss": 0.2124,
+      "step": 8411
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022022415940224158,
+      "loss": 0.2679,
+      "step": 8412
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00022021170610211707,
+      "loss": 0.2073,
+      "step": 8413
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0002201992528019925,
+      "loss": 0.2243,
+      "step": 8414
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022018679950186796,
+      "loss": 0.3029,
+      "step": 8415
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022017434620174344,
+      "loss": 0.341,
+      "step": 8416
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002201618929016189,
+      "loss": 0.1452,
+      "step": 8417
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002201494396014944,
+      "loss": 0.2191,
+      "step": 8418
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022013698630136982,
+      "loss": 0.2882,
+      "step": 8419
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002201245330012453,
+      "loss": 0.2484,
+      "step": 8420
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022011207970112077,
+      "loss": 0.6207,
+      "step": 8421
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022009962640099625,
+      "loss": 0.3441,
+      "step": 8422
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002200871731008717,
+      "loss": 0.1865,
+      "step": 8423
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002200747198007472,
+      "loss": 0.2659,
+      "step": 8424
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022006226650062263,
+      "loss": 0.2756,
+      "step": 8425
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022004981320049812,
+      "loss": 0.1786,
+      "step": 8426
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022003735990037358,
+      "loss": 0.3038,
+      "step": 8427
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022002490660024906,
+      "loss": 0.3735,
+      "step": 8428
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00022001245330012452,
+      "loss": 0.2768,
+      "step": 8429
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00021999999999999995,
+      "loss": 0.1833,
+      "step": 8430
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00021998754669987544,
+      "loss": 0.2569,
+      "step": 8431
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002199750933997509,
+      "loss": 0.1784,
+      "step": 8432
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00021996264009962639,
+      "loss": 0.3134,
+      "step": 8433
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00021995018679950184,
+      "loss": 0.2728,
+      "step": 8434
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00021993773349937733,
+      "loss": 0.278,
+      "step": 8435
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00021992528019925276,
+      "loss": 0.1946,
+      "step": 8436
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00021991282689912825,
+      "loss": 0.255,
+      "step": 8437
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002199003735990037,
+      "loss": 0.3175,
+      "step": 8438
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002198879202988792,
+      "loss": 0.1955,
+      "step": 8439
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00021987546699875465,
+      "loss": 0.37,
+      "step": 8440
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021986301369863014,
+      "loss": 0.2819,
+      "step": 8441
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021985056039850557,
+      "loss": 0.2316,
+      "step": 8442
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021983810709838106,
+      "loss": 0.1639,
+      "step": 8443
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021982565379825652,
+      "loss": 0.1003,
+      "step": 8444
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021981320049813198,
+      "loss": 0.344,
+      "step": 8445
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021980074719800746,
+      "loss": 0.3945,
+      "step": 8446
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0002197882938978829,
+      "loss": 0.1947,
+      "step": 8447
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021977584059775838,
+      "loss": 0.3475,
+      "step": 8448
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021976338729763384,
+      "loss": 0.2103,
+      "step": 8449
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021975093399750933,
+      "loss": 0.2204,
+      "step": 8450
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021973848069738479,
+      "loss": 0.23,
+      "step": 8451
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021972602739726027,
+      "loss": 0.292,
+      "step": 8452
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0002197135740971357,
+      "loss": 0.4504,
+      "step": 8453
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0002197011207970112,
+      "loss": 0.2533,
+      "step": 8454
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021968866749688665,
+      "loss": 0.2068,
+      "step": 8455
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021967621419676214,
+      "loss": 0.1729,
+      "step": 8456
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0002196637608966376,
+      "loss": 0.2197,
+      "step": 8457
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021965130759651308,
+      "loss": 0.2892,
+      "step": 8458
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0002196388542963885,
+      "loss": 0.3787,
+      "step": 8459
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.000219626400996264,
+      "loss": 0.2202,
+      "step": 8460
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021961394769613946,
+      "loss": 0.1927,
+      "step": 8461
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021960149439601492,
+      "loss": 0.3687,
+      "step": 8462
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0002195890410958904,
+      "loss": 0.153,
+      "step": 8463
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021957658779576584,
+      "loss": 0.1856,
+      "step": 8464
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021956413449564132,
+      "loss": 0.3128,
+      "step": 8465
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00021955168119551678,
+      "loss": 0.2238,
+      "step": 8466
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021953922789539227,
+      "loss": 0.1937,
+      "step": 8467
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021952677459526773,
+      "loss": 0.4005,
+      "step": 8468
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021951432129514321,
+      "loss": 0.2207,
+      "step": 8469
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021950186799501865,
+      "loss": 0.2474,
+      "step": 8470
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021948941469489413,
+      "loss": 0.3336,
+      "step": 8471
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0002194769613947696,
+      "loss": 0.1988,
+      "step": 8472
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021946450809464508,
+      "loss": 0.2129,
+      "step": 8473
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021945205479452054,
+      "loss": 0.3303,
+      "step": 8474
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021943960149439602,
+      "loss": 0.1418,
+      "step": 8475
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021942714819427145,
+      "loss": 0.1901,
+      "step": 8476
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021941469489414691,
+      "loss": 0.4049,
+      "step": 8477
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0002194022415940224,
+      "loss": 0.2423,
+      "step": 8478
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021938978829389786,
+      "loss": 0.1895,
+      "step": 8479
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021937733499377335,
+      "loss": 0.1728,
+      "step": 8480
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021936488169364878,
+      "loss": 0.2406,
+      "step": 8481
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021935242839352426,
+      "loss": 0.2302,
+      "step": 8482
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021933997509339972,
+      "loss": 0.2441,
+      "step": 8483
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0002193275217932752,
+      "loss": 0.3038,
+      "step": 8484
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021931506849315067,
+      "loss": 0.2193,
+      "step": 8485
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021930261519302615,
+      "loss": 0.2483,
+      "step": 8486
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0002192901618929016,
+      "loss": 0.2319,
+      "step": 8487
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021927770859277707,
+      "loss": 0.3074,
+      "step": 8488
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021926525529265253,
+      "loss": 0.3874,
+      "step": 8489
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021925280199252802,
+      "loss": 0.4523,
+      "step": 8490
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00021924034869240348,
+      "loss": 0.2015,
+      "step": 8491
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0002192278953922789,
+      "loss": 0.2057,
+      "step": 8492
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002192154420921544,
+      "loss": 0.4853,
+      "step": 8493
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021920298879202986,
+      "loss": 0.1981,
+      "step": 8494
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021919053549190534,
+      "loss": 0.1905,
+      "step": 8495
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002191780821917808,
+      "loss": 0.448,
+      "step": 8496
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002191656288916563,
+      "loss": 0.2346,
+      "step": 8497
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021915317559153172,
+      "loss": 0.2207,
+      "step": 8498
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002191407222914072,
+      "loss": 0.202,
+      "step": 8499
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021912826899128266,
+      "loss": 0.2933,
+      "step": 8500
+    },
+    {
+      "epoch": 3.26,
+      "eval_cer": 0.1103732173166592,
+      "eval_loss": 0.3672265410423279,
+      "eval_runtime": 47.2073,
+      "eval_samples_per_second": 37.134,
+      "eval_steps_per_second": 4.66,
+      "eval_wer": 0.19154183825858714,
+      "step": 8500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021911581569115815,
+      "loss": 0.1842,
+      "step": 8501
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002191033623910336,
+      "loss": 0.4953,
+      "step": 8502
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002190909090909091,
+      "loss": 0.2245,
+      "step": 8503
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021907845579078453,
+      "loss": 0.2431,
+      "step": 8504
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021906600249066001,
+      "loss": 0.2794,
+      "step": 8505
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021905354919053547,
+      "loss": 0.2049,
+      "step": 8506
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021904109589041096,
+      "loss": 0.2412,
+      "step": 8507
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021902864259028642,
+      "loss": 0.4052,
+      "step": 8508
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021901618929016185,
+      "loss": 0.3845,
+      "step": 8509
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021900373599003734,
+      "loss": 0.2282,
+      "step": 8510
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002189912826899128,
+      "loss": 0.1966,
+      "step": 8511
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021897882938978828,
+      "loss": 0.2395,
+      "step": 8512
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021896637608966374,
+      "loss": 0.243,
+      "step": 8513
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021895392278953923,
+      "loss": 0.1718,
+      "step": 8514
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021894146948941466,
+      "loss": 0.4462,
+      "step": 8515
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00021892901618929015,
+      "loss": 0.1699,
+      "step": 8516
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002189165628891656,
+      "loss": 0.2173,
+      "step": 8517
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0002189041095890411,
+      "loss": 0.1823,
+      "step": 8518
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021889165628891655,
+      "loss": 0.1618,
+      "step": 8519
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.000218879202988792,
+      "loss": 0.2111,
+      "step": 8520
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021886674968866747,
+      "loss": 0.2627,
+      "step": 8521
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021885429638854296,
+      "loss": 0.2212,
+      "step": 8522
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021884184308841842,
+      "loss": 0.1944,
+      "step": 8523
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021882938978829387,
+      "loss": 0.2117,
+      "step": 8524
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021881693648816936,
+      "loss": 0.1594,
+      "step": 8525
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0002188044831880448,
+      "loss": 0.1732,
+      "step": 8526
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021879202988792028,
+      "loss": 0.3142,
+      "step": 8527
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021877957658779574,
+      "loss": 0.2364,
+      "step": 8528
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021876712328767122,
+      "loss": 0.2526,
+      "step": 8529
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021875466998754668,
+      "loss": 0.1772,
+      "step": 8530
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021874221668742214,
+      "loss": 0.2917,
+      "step": 8531
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0002187297633872976,
+      "loss": 0.304,
+      "step": 8532
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0002187173100871731,
+      "loss": 0.2802,
+      "step": 8533
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021870485678704855,
+      "loss": 0.3494,
+      "step": 8534
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021869240348692403,
+      "loss": 0.1722,
+      "step": 8535
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0002186799501867995,
+      "loss": 0.1516,
+      "step": 8536
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021866749688667495,
+      "loss": 0.2151,
+      "step": 8537
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0002186550435865504,
+      "loss": 0.2376,
+      "step": 8538
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021864259028642587,
+      "loss": 0.3769,
+      "step": 8539
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021863013698630136,
+      "loss": 0.528,
+      "step": 8540
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021861768368617682,
+      "loss": 0.2383,
+      "step": 8541
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021860523038605227,
+      "loss": 0.3547,
+      "step": 8542
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021859277708592773,
+      "loss": 0.2058,
+      "step": 8543
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00021858032378580322,
+      "loss": 0.1988,
+      "step": 8544
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021856787048567868,
+      "loss": 0.3115,
+      "step": 8545
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021855541718555417,
+      "loss": 0.3424,
+      "step": 8546
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021854296388542963,
+      "loss": 0.1816,
+      "step": 8547
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021853051058530508,
+      "loss": 0.2655,
+      "step": 8548
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021851805728518054,
+      "loss": 0.2842,
+      "step": 8549
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021850560398505603,
+      "loss": 0.2902,
+      "step": 8550
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0002184931506849315,
+      "loss": 0.1981,
+      "step": 8551
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021848069738480698,
+      "loss": 0.3517,
+      "step": 8552
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0002184682440846824,
+      "loss": 0.2823,
+      "step": 8553
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0002184557907845579,
+      "loss": 0.2792,
+      "step": 8554
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021844333748443335,
+      "loss": 0.257,
+      "step": 8555
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0002184308841843088,
+      "loss": 0.2518,
+      "step": 8556
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0002184184308841843,
+      "loss": 0.1679,
+      "step": 8557
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021840597758405976,
+      "loss": 0.2036,
+      "step": 8558
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021839352428393522,
+      "loss": 0.4844,
+      "step": 8559
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021838107098381068,
+      "loss": 0.2001,
+      "step": 8560
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021836861768368616,
+      "loss": 0.1879,
+      "step": 8561
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021835616438356162,
+      "loss": 0.1486,
+      "step": 8562
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0002183437110834371,
+      "loss": 0.3017,
+      "step": 8563
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021833125778331254,
+      "loss": 0.2634,
+      "step": 8564
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021831880448318803,
+      "loss": 0.4032,
+      "step": 8565
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021830635118306348,
+      "loss": 0.1827,
+      "step": 8566
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021829389788293897,
+      "loss": 0.2407,
+      "step": 8567
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021828144458281443,
+      "loss": 0.235,
+      "step": 8568
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021826899128268992,
+      "loss": 0.1498,
+      "step": 8569
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00021825653798256535,
+      "loss": 0.1817,
+      "step": 8570
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002182440846824408,
+      "loss": 0.236,
+      "step": 8571
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002182316313823163,
+      "loss": 0.1789,
+      "step": 8572
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021821917808219175,
+      "loss": 0.2091,
+      "step": 8573
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021820672478206724,
+      "loss": 0.2465,
+      "step": 8574
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021819427148194267,
+      "loss": 0.1523,
+      "step": 8575
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021818181818181816,
+      "loss": 0.2667,
+      "step": 8576
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021816936488169362,
+      "loss": 0.3507,
+      "step": 8577
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002181569115815691,
+      "loss": 0.2684,
+      "step": 8578
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021814445828144456,
+      "loss": 0.2122,
+      "step": 8579
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021813200498132005,
+      "loss": 0.1918,
+      "step": 8580
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021811955168119548,
+      "loss": 0.2791,
+      "step": 8581
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021810709838107097,
+      "loss": 0.2038,
+      "step": 8582
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021809464508094643,
+      "loss": 0.175,
+      "step": 8583
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002180821917808219,
+      "loss": 0.2619,
+      "step": 8584
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021806973848069737,
+      "loss": 0.5643,
+      "step": 8585
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002180572851805728,
+      "loss": 0.2073,
+      "step": 8586
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002180448318804483,
+      "loss": 0.1579,
+      "step": 8587
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021803237858032375,
+      "loss": 0.1919,
+      "step": 8588
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021801992528019924,
+      "loss": 0.1953,
+      "step": 8589
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002180074719800747,
+      "loss": 0.2802,
+      "step": 8590
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021799501867995018,
+      "loss": 0.2114,
+      "step": 8591
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002179825653798256,
+      "loss": 0.2762,
+      "step": 8592
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002179701120797011,
+      "loss": 0.3098,
+      "step": 8593
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021795765877957656,
+      "loss": 0.2336,
+      "step": 8594
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00021794520547945204,
+      "loss": 0.1744,
+      "step": 8595
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.0002179327521793275,
+      "loss": 0.2636,
+      "step": 8596
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.000217920298879203,
+      "loss": 0.2988,
+      "step": 8597
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021790784557907842,
+      "loss": 0.3212,
+      "step": 8598
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0002178953922789539,
+      "loss": 0.2205,
+      "step": 8599
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021788293897882937,
+      "loss": 0.2249,
+      "step": 8600
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021787048567870485,
+      "loss": 0.316,
+      "step": 8601
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0002178580323785803,
+      "loss": 0.387,
+      "step": 8602
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021784557907845575,
+      "loss": 0.2951,
+      "step": 8603
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021783312577833123,
+      "loss": 0.2893,
+      "step": 8604
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0002178206724782067,
+      "loss": 0.207,
+      "step": 8605
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021780821917808218,
+      "loss": 0.2652,
+      "step": 8606
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021779576587795764,
+      "loss": 0.2592,
+      "step": 8607
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021778331257783312,
+      "loss": 0.3038,
+      "step": 8608
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021777085927770855,
+      "loss": 0.5349,
+      "step": 8609
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021775840597758404,
+      "loss": 0.2064,
+      "step": 8610
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0002177459526774595,
+      "loss": 0.2376,
+      "step": 8611
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021773349937733499,
+      "loss": 0.3495,
+      "step": 8612
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021772104607721045,
+      "loss": 0.299,
+      "step": 8613
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021770859277708593,
+      "loss": 0.2552,
+      "step": 8614
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021769613947696136,
+      "loss": 0.2845,
+      "step": 8615
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021768368617683685,
+      "loss": 0.2312,
+      "step": 8616
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0002176712328767123,
+      "loss": 0.1521,
+      "step": 8617
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021765877957658777,
+      "loss": 0.2037,
+      "step": 8618
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021764632627646325,
+      "loss": 0.187,
+      "step": 8619
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0002176338729763387,
+      "loss": 0.2344,
+      "step": 8620
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021762141967621417,
+      "loss": 0.3532,
+      "step": 8621
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00021760896637608963,
+      "loss": 0.2058,
+      "step": 8622
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021759651307596512,
+      "loss": 0.2164,
+      "step": 8623
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021758405977584058,
+      "loss": 0.2183,
+      "step": 8624
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021757160647571606,
+      "loss": 0.3099,
+      "step": 8625
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0002175591531755915,
+      "loss": 0.195,
+      "step": 8626
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021754669987546698,
+      "loss": 0.3271,
+      "step": 8627
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021753424657534244,
+      "loss": 0.2095,
+      "step": 8628
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021752179327521793,
+      "loss": 0.4,
+      "step": 8629
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0002175093399750934,
+      "loss": 0.1782,
+      "step": 8630
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021749688667496887,
+      "loss": 0.464,
+      "step": 8631
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0002174844333748443,
+      "loss": 0.2177,
+      "step": 8632
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0002174719800747198,
+      "loss": 0.2878,
+      "step": 8633
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021745952677459525,
+      "loss": 0.4585,
+      "step": 8634
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0002174470734744707,
+      "loss": 0.2657,
+      "step": 8635
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0002174346201743462,
+      "loss": 0.3444,
+      "step": 8636
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021742216687422163,
+      "loss": 0.2879,
+      "step": 8637
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021740971357409711,
+      "loss": 0.2082,
+      "step": 8638
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021739726027397257,
+      "loss": 0.1802,
+      "step": 8639
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021738480697384806,
+      "loss": 0.3849,
+      "step": 8640
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021737235367372352,
+      "loss": 0.2238,
+      "step": 8641
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.000217359900373599,
+      "loss": 0.3133,
+      "step": 8642
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021734744707347444,
+      "loss": 0.2062,
+      "step": 8643
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021733499377334992,
+      "loss": 0.2293,
+      "step": 8644
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021732254047322538,
+      "loss": 0.1844,
+      "step": 8645
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021731008717310087,
+      "loss": 0.4825,
+      "step": 8646
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021729763387297633,
+      "loss": 0.2202,
+      "step": 8647
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00021728518057285181,
+      "loss": 0.1955,
+      "step": 8648
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021727272727272725,
+      "loss": 0.2627,
+      "step": 8649
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0002172602739726027,
+      "loss": 0.1619,
+      "step": 8650
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0002172478206724782,
+      "loss": 0.1668,
+      "step": 8651
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021723536737235365,
+      "loss": 0.3964,
+      "step": 8652
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021722291407222914,
+      "loss": 0.2981,
+      "step": 8653
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021721046077210457,
+      "loss": 0.3717,
+      "step": 8654
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021719800747198006,
+      "loss": 0.3566,
+      "step": 8655
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021718555417185551,
+      "loss": 0.1983,
+      "step": 8656
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.000217173100871731,
+      "loss": 0.2046,
+      "step": 8657
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021716064757160646,
+      "loss": 0.3235,
+      "step": 8658
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021714819427148195,
+      "loss": 0.5152,
+      "step": 8659
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021713574097135738,
+      "loss": 0.2205,
+      "step": 8660
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021712328767123286,
+      "loss": 0.1871,
+      "step": 8661
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021711083437110832,
+      "loss": 0.1536,
+      "step": 8662
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0002170983810709838,
+      "loss": 0.3188,
+      "step": 8663
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021708592777085927,
+      "loss": 0.1833,
+      "step": 8664
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0002170734744707347,
+      "loss": 0.3596,
+      "step": 8665
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0002170610211706102,
+      "loss": 0.2045,
+      "step": 8666
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021704856787048565,
+      "loss": 0.3284,
+      "step": 8667
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021703611457036113,
+      "loss": 0.3162,
+      "step": 8668
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0002170236612702366,
+      "loss": 0.2082,
+      "step": 8669
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021701120797011208,
+      "loss": 0.3595,
+      "step": 8670
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0002169987546699875,
+      "loss": 0.3661,
+      "step": 8671
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.000216986301369863,
+      "loss": 0.2122,
+      "step": 8672
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021697384806973846,
+      "loss": 0.194,
+      "step": 8673
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00021696139476961394,
+      "loss": 0.2122,
+      "step": 8674
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002169489414694894,
+      "loss": 0.1907,
+      "step": 8675
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002169364881693649,
+      "loss": 0.2894,
+      "step": 8676
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021692403486924032,
+      "loss": 0.458,
+      "step": 8677
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002169115815691158,
+      "loss": 0.3283,
+      "step": 8678
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021689912826899127,
+      "loss": 0.3423,
+      "step": 8679
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021688667496886675,
+      "loss": 0.3183,
+      "step": 8680
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002168742216687422,
+      "loss": 0.1985,
+      "step": 8681
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021686176836861764,
+      "loss": 0.1864,
+      "step": 8682
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021684931506849313,
+      "loss": 0.2847,
+      "step": 8683
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002168368617683686,
+      "loss": 0.4104,
+      "step": 8684
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021682440846824407,
+      "loss": 0.2402,
+      "step": 8685
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021681195516811953,
+      "loss": 0.2992,
+      "step": 8686
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021679950186799502,
+      "loss": 0.232,
+      "step": 8687
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021678704856787045,
+      "loss": 0.2008,
+      "step": 8688
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021677459526774594,
+      "loss": 0.3071,
+      "step": 8689
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002167621419676214,
+      "loss": 0.4209,
+      "step": 8690
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021674968866749688,
+      "loss": 0.2216,
+      "step": 8691
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021673723536737234,
+      "loss": 0.2145,
+      "step": 8692
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002167247820672478,
+      "loss": 0.302,
+      "step": 8693
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021671232876712326,
+      "loss": 0.1736,
+      "step": 8694
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021669987546699875,
+      "loss": 0.2576,
+      "step": 8695
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0002166874221668742,
+      "loss": 0.3024,
+      "step": 8696
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021667496886674967,
+      "loss": 0.268,
+      "step": 8697
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021666251556662515,
+      "loss": 0.2048,
+      "step": 8698
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021665006226650058,
+      "loss": 0.2338,
+      "step": 8699
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021663760896637607,
+      "loss": 0.2098,
+      "step": 8700
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00021662515566625153,
+      "loss": 0.3228,
+      "step": 8701
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021661270236612702,
+      "loss": 0.3748,
+      "step": 8702
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021660024906600248,
+      "loss": 0.3123,
+      "step": 8703
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021658779576587793,
+      "loss": 0.2026,
+      "step": 8704
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0002165753424657534,
+      "loss": 0.2624,
+      "step": 8705
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021656288916562888,
+      "loss": 0.2422,
+      "step": 8706
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021655043586550434,
+      "loss": 0.2138,
+      "step": 8707
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021653798256537983,
+      "loss": 0.2284,
+      "step": 8708
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021652552926525528,
+      "loss": 0.243,
+      "step": 8709
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021651307596513074,
+      "loss": 0.1656,
+      "step": 8710
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0002165006226650062,
+      "loss": 0.3716,
+      "step": 8711
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021648816936488166,
+      "loss": 0.2624,
+      "step": 8712
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021647571606475715,
+      "loss": 0.2107,
+      "step": 8713
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0002164632627646326,
+      "loss": 0.188,
+      "step": 8714
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021645080946450807,
+      "loss": 0.3393,
+      "step": 8715
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021643835616438353,
+      "loss": 0.2416,
+      "step": 8716
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.000216425902864259,
+      "loss": 0.2293,
+      "step": 8717
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021641344956413447,
+      "loss": 0.2349,
+      "step": 8718
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021640099626400996,
+      "loss": 0.3163,
+      "step": 8719
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0002163885429638854,
+      "loss": 0.2776,
+      "step": 8720
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021637608966376088,
+      "loss": 0.4155,
+      "step": 8721
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021636363636363633,
+      "loss": 0.2247,
+      "step": 8722
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021635118306351182,
+      "loss": 0.1712,
+      "step": 8723
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021633872976338728,
+      "loss": 0.151,
+      "step": 8724
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021632627646326277,
+      "loss": 0.1996,
+      "step": 8725
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0002163138231631382,
+      "loss": 0.2338,
+      "step": 8726
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00021630136986301368,
+      "loss": 0.2986,
+      "step": 8727
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021628891656288914,
+      "loss": 0.291,
+      "step": 8728
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0002162764632627646,
+      "loss": 0.3158,
+      "step": 8729
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0002162640099626401,
+      "loss": 0.2414,
+      "step": 8730
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021625155666251552,
+      "loss": 0.2413,
+      "step": 8731
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.000216239103362391,
+      "loss": 0.1989,
+      "step": 8732
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021622665006226647,
+      "loss": 0.2692,
+      "step": 8733
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021621419676214195,
+      "loss": 0.265,
+      "step": 8734
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0002162017434620174,
+      "loss": 0.2506,
+      "step": 8735
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0002161892901618929,
+      "loss": 0.2592,
+      "step": 8736
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021617683686176833,
+      "loss": 0.1562,
+      "step": 8737
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021616438356164382,
+      "loss": 0.2273,
+      "step": 8738
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021615193026151928,
+      "loss": 0.2016,
+      "step": 8739
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021613947696139476,
+      "loss": 0.3425,
+      "step": 8740
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021612702366127022,
+      "loss": 0.1966,
+      "step": 8741
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0002161145703611457,
+      "loss": 0.2334,
+      "step": 8742
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021610211706102114,
+      "loss": 0.2695,
+      "step": 8743
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0002160896637608966,
+      "loss": 0.2697,
+      "step": 8744
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021607721046077209,
+      "loss": 0.1803,
+      "step": 8745
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021606475716064754,
+      "loss": 0.3113,
+      "step": 8746
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021605230386052303,
+      "loss": 0.401,
+      "step": 8747
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021603985056039846,
+      "loss": 0.2695,
+      "step": 8748
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021602739726027395,
+      "loss": 0.2087,
+      "step": 8749
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0002160149439601494,
+      "loss": 0.1985,
+      "step": 8750
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0002160024906600249,
+      "loss": 0.3695,
+      "step": 8751
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021599003735990035,
+      "loss": 0.3946,
+      "step": 8752
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00021597758405977584,
+      "loss": 0.2302,
+      "step": 8753
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021596513075965127,
+      "loss": 0.3029,
+      "step": 8754
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021595267745952676,
+      "loss": 0.1919,
+      "step": 8755
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021594022415940222,
+      "loss": 0.1877,
+      "step": 8756
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0002159277708592777,
+      "loss": 0.2772,
+      "step": 8757
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021591531755915316,
+      "loss": 0.3089,
+      "step": 8758
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0002159028642590286,
+      "loss": 0.247,
+      "step": 8759
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021589041095890408,
+      "loss": 0.231,
+      "step": 8760
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021587795765877954,
+      "loss": 0.2231,
+      "step": 8761
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021586550435865503,
+      "loss": 0.1747,
+      "step": 8762
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021585305105853049,
+      "loss": 0.2175,
+      "step": 8763
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021584059775840597,
+      "loss": 0.2052,
+      "step": 8764
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0002158281444582814,
+      "loss": 0.3121,
+      "step": 8765
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0002158156911581569,
+      "loss": 0.1882,
+      "step": 8766
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021580323785803235,
+      "loss": 0.1616,
+      "step": 8767
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021579078455790784,
+      "loss": 0.1772,
+      "step": 8768
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0002157783312577833,
+      "loss": 0.3496,
+      "step": 8769
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021576587795765878,
+      "loss": 0.3153,
+      "step": 8770
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021575342465753421,
+      "loss": 0.563,
+      "step": 8771
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0002157409713574097,
+      "loss": 0.2513,
+      "step": 8772
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021572851805728516,
+      "loss": 0.309,
+      "step": 8773
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021571606475716065,
+      "loss": 0.1915,
+      "step": 8774
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0002157036114570361,
+      "loss": 0.1807,
+      "step": 8775
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021569115815691154,
+      "loss": 0.1895,
+      "step": 8776
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021567870485678702,
+      "loss": 0.2263,
+      "step": 8777
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021566625155666248,
+      "loss": 0.2712,
+      "step": 8778
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00021565379825653797,
+      "loss": 0.2999,
+      "step": 8779
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021564134495641343,
+      "loss": 0.2705,
+      "step": 8780
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021562889165628891,
+      "loss": 0.24,
+      "step": 8781
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021561643835616435,
+      "loss": 0.2602,
+      "step": 8782
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021560398505603983,
+      "loss": 0.3987,
+      "step": 8783
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0002155915317559153,
+      "loss": 0.328,
+      "step": 8784
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021557907845579078,
+      "loss": 0.2522,
+      "step": 8785
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021556662515566624,
+      "loss": 0.3494,
+      "step": 8786
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021555417185554172,
+      "loss": 0.1429,
+      "step": 8787
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021554171855541715,
+      "loss": 0.2173,
+      "step": 8788
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021552926525529264,
+      "loss": 0.1926,
+      "step": 8789
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0002155168119551681,
+      "loss": 0.4848,
+      "step": 8790
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021550435865504356,
+      "loss": 0.2294,
+      "step": 8791
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021549190535491905,
+      "loss": 0.2906,
+      "step": 8792
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021547945205479448,
+      "loss": 0.224,
+      "step": 8793
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021546699875466996,
+      "loss": 0.2117,
+      "step": 8794
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021545454545454542,
+      "loss": 0.1881,
+      "step": 8795
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0002154420921544209,
+      "loss": 0.2406,
+      "step": 8796
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021542963885429637,
+      "loss": 0.2165,
+      "step": 8797
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021541718555417186,
+      "loss": 0.2291,
+      "step": 8798
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0002154047322540473,
+      "loss": 0.1983,
+      "step": 8799
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021539227895392277,
+      "loss": 0.3171,
+      "step": 8800
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021537982565379823,
+      "loss": 0.2693,
+      "step": 8801
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021536737235367372,
+      "loss": 0.4256,
+      "step": 8802
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021535491905354918,
+      "loss": 0.2277,
+      "step": 8803
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00021534246575342466,
+      "loss": 0.2379,
+      "step": 8804
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0002153300124533001,
+      "loss": 0.2159,
+      "step": 8805
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021531755915317556,
+      "loss": 0.2239,
+      "step": 8806
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021530510585305104,
+      "loss": 0.2284,
+      "step": 8807
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0002152926525529265,
+      "loss": 0.3759,
+      "step": 8808
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.000215280199252802,
+      "loss": 0.4228,
+      "step": 8809
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021526774595267742,
+      "loss": 0.2405,
+      "step": 8810
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0002152552926525529,
+      "loss": 0.1603,
+      "step": 8811
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021524283935242836,
+      "loss": 0.2001,
+      "step": 8812
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021523038605230385,
+      "loss": 0.2789,
+      "step": 8813
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0002152179327521793,
+      "loss": 0.3469,
+      "step": 8814
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0002152054794520548,
+      "loss": 0.3283,
+      "step": 8815
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021519302615193023,
+      "loss": 0.2268,
+      "step": 8816
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021518057285180571,
+      "loss": 0.1969,
+      "step": 8817
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021516811955168117,
+      "loss": 0.1861,
+      "step": 8818
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021515566625155666,
+      "loss": 0.1978,
+      "step": 8819
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021514321295143212,
+      "loss": 0.1601,
+      "step": 8820
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0002151307596513076,
+      "loss": 0.2735,
+      "step": 8821
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021511830635118304,
+      "loss": 0.2586,
+      "step": 8822
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0002151058530510585,
+      "loss": 0.312,
+      "step": 8823
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021509339975093398,
+      "loss": 0.1968,
+      "step": 8824
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021508094645080944,
+      "loss": 0.1939,
+      "step": 8825
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021506849315068493,
+      "loss": 0.2991,
+      "step": 8826
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021505603985056036,
+      "loss": 0.5119,
+      "step": 8827
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021504358655043585,
+      "loss": 0.3488,
+      "step": 8828
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0002150311332503113,
+      "loss": 0.1919,
+      "step": 8829
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0002150186799501868,
+      "loss": 0.2953,
+      "step": 8830
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00021500622665006225,
+      "loss": 0.1669,
+      "step": 8831
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021499377334993774,
+      "loss": 0.2823,
+      "step": 8832
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021498132004981317,
+      "loss": 0.3389,
+      "step": 8833
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021496886674968866,
+      "loss": 0.2853,
+      "step": 8834
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021495641344956412,
+      "loss": 0.2433,
+      "step": 8835
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0002149439601494396,
+      "loss": 0.21,
+      "step": 8836
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021493150684931506,
+      "loss": 0.172,
+      "step": 8837
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0002149190535491905,
+      "loss": 0.2551,
+      "step": 8838
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021490660024906598,
+      "loss": 0.2897,
+      "step": 8839
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021489414694894144,
+      "loss": 0.4002,
+      "step": 8840
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021488169364881692,
+      "loss": 0.5202,
+      "step": 8841
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021486924034869238,
+      "loss": 0.1721,
+      "step": 8842
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021485678704856787,
+      "loss": 0.216,
+      "step": 8843
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0002148443337484433,
+      "loss": 0.1994,
+      "step": 8844
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0002148318804483188,
+      "loss": 0.2618,
+      "step": 8845
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021481942714819425,
+      "loss": 0.3474,
+      "step": 8846
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021480697384806973,
+      "loss": 0.2313,
+      "step": 8847
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0002147945205479452,
+      "loss": 0.2231,
+      "step": 8848
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021478206724782068,
+      "loss": 0.2871,
+      "step": 8849
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0002147696139476961,
+      "loss": 0.1939,
+      "step": 8850
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0002147571606475716,
+      "loss": 0.2186,
+      "step": 8851
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021474470734744706,
+      "loss": 0.3477,
+      "step": 8852
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021473225404732252,
+      "loss": 0.2423,
+      "step": 8853
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.000214719800747198,
+      "loss": 0.4379,
+      "step": 8854
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021470734744707343,
+      "loss": 0.2107,
+      "step": 8855
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021469489414694892,
+      "loss": 0.2103,
+      "step": 8856
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00021468244084682438,
+      "loss": 0.1882,
+      "step": 8857
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021466998754669987,
+      "loss": 0.267,
+      "step": 8858
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021465753424657533,
+      "loss": 0.2113,
+      "step": 8859
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0002146450809464508,
+      "loss": 0.2424,
+      "step": 8860
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021463262764632624,
+      "loss": 0.2668,
+      "step": 8861
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021462017434620173,
+      "loss": 0.1876,
+      "step": 8862
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0002146077210460772,
+      "loss": 0.2638,
+      "step": 8863
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021459526774595268,
+      "loss": 0.2219,
+      "step": 8864
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021458281444582813,
+      "loss": 0.4039,
+      "step": 8865
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0002145703611457036,
+      "loss": 0.2319,
+      "step": 8866
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021455790784557905,
+      "loss": 0.2475,
+      "step": 8867
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021454545454545454,
+      "loss": 0.1845,
+      "step": 8868
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021453300124533,
+      "loss": 0.1999,
+      "step": 8869
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021452054794520546,
+      "loss": 0.2385,
+      "step": 8870
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021450809464508094,
+      "loss": 0.3849,
+      "step": 8871
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021449564134495638,
+      "loss": 0.2458,
+      "step": 8872
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021448318804483186,
+      "loss": 0.1915,
+      "step": 8873
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021447073474470732,
+      "loss": 0.1719,
+      "step": 8874
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0002144582814445828,
+      "loss": 0.2792,
+      "step": 8875
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021444582814445827,
+      "loss": 0.2541,
+      "step": 8876
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021443337484433373,
+      "loss": 0.4272,
+      "step": 8877
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021442092154420918,
+      "loss": 0.2289,
+      "step": 8878
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021440846824408467,
+      "loss": 0.1831,
+      "step": 8879
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021439601494396013,
+      "loss": 0.1972,
+      "step": 8880
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021438356164383562,
+      "loss": 0.1368,
+      "step": 8881
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021437110834371105,
+      "loss": 0.2189,
+      "step": 8882
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00021435865504358653,
+      "loss": 0.2618,
+      "step": 8883
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.000214346201743462,
+      "loss": 0.4352,
+      "step": 8884
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021433374844333745,
+      "loss": 0.2566,
+      "step": 8885
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021432129514321294,
+      "loss": 0.2467,
+      "step": 8886
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0002143088418430884,
+      "loss": 0.2046,
+      "step": 8887
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021429638854296386,
+      "loss": 0.1728,
+      "step": 8888
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021428393524283932,
+      "loss": 0.3244,
+      "step": 8889
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0002142714819427148,
+      "loss": 0.4173,
+      "step": 8890
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021425902864259026,
+      "loss": 0.2691,
+      "step": 8891
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021424657534246575,
+      "loss": 0.2705,
+      "step": 8892
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021423412204234118,
+      "loss": 0.169,
+      "step": 8893
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021422166874221667,
+      "loss": 0.2024,
+      "step": 8894
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021420921544209213,
+      "loss": 0.2527,
+      "step": 8895
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0002141967621419676,
+      "loss": 0.3725,
+      "step": 8896
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021418430884184307,
+      "loss": 0.1868,
+      "step": 8897
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021417185554171856,
+      "loss": 0.1987,
+      "step": 8898
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.000214159402241594,
+      "loss": 0.2016,
+      "step": 8899
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021414694894146948,
+      "loss": 0.3593,
+      "step": 8900
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021413449564134494,
+      "loss": 0.2241,
+      "step": 8901
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0002141220423412204,
+      "loss": 0.2756,
+      "step": 8902
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021410958904109588,
+      "loss": 0.2167,
+      "step": 8903
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0002140971357409713,
+      "loss": 0.2357,
+      "step": 8904
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0002140846824408468,
+      "loss": 0.2695,
+      "step": 8905
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021407222914072226,
+      "loss": 0.1936,
+      "step": 8906
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00021405977584059774,
+      "loss": 0.1588,
+      "step": 8907
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0002140473225404732,
+      "loss": 0.2886,
+      "step": 8908
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0002140348692403487,
+      "loss": 0.4524,
+      "step": 8909
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021402241594022412,
+      "loss": 0.2724,
+      "step": 8910
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0002140099626400996,
+      "loss": 0.1872,
+      "step": 8911
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021399750933997507,
+      "loss": 0.2736,
+      "step": 8912
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021398505603985055,
+      "loss": 0.2765,
+      "step": 8913
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.000213972602739726,
+      "loss": 0.2926,
+      "step": 8914
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0002139601494396015,
+      "loss": 0.4555,
+      "step": 8915
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021394769613947693,
+      "loss": 0.1983,
+      "step": 8916
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0002139352428393524,
+      "loss": 0.1914,
+      "step": 8917
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021392278953922788,
+      "loss": 0.1816,
+      "step": 8918
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021391033623910334,
+      "loss": 0.2267,
+      "step": 8919
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021389788293897882,
+      "loss": 0.2899,
+      "step": 8920
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021388542963885425,
+      "loss": 0.4256,
+      "step": 8921
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021387297633872974,
+      "loss": 0.2104,
+      "step": 8922
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0002138605230386052,
+      "loss": 0.2039,
+      "step": 8923
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021384806973848069,
+      "loss": 0.2794,
+      "step": 8924
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021383561643835615,
+      "loss": 0.1874,
+      "step": 8925
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021382316313823163,
+      "loss": 0.1963,
+      "step": 8926
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021381070983810706,
+      "loss": 0.3814,
+      "step": 8927
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021379825653798255,
+      "loss": 0.22,
+      "step": 8928
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.000213785803237858,
+      "loss": 0.2778,
+      "step": 8929
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0002137733499377335,
+      "loss": 0.2387,
+      "step": 8930
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021376089663760895,
+      "loss": 0.2105,
+      "step": 8931
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0002137484433374844,
+      "loss": 0.2583,
+      "step": 8932
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021373599003735987,
+      "loss": 0.2369,
+      "step": 8933
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021372353673723533,
+      "loss": 0.2686,
+      "step": 8934
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00021371108343711082,
+      "loss": 0.2281,
+      "step": 8935
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021369863013698628,
+      "loss": 0.2343,
+      "step": 8936
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021368617683686176,
+      "loss": 0.1613,
+      "step": 8937
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0002136737235367372,
+      "loss": 0.2305,
+      "step": 8938
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021366127023661268,
+      "loss": 0.4281,
+      "step": 8939
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021364881693648814,
+      "loss": 0.3673,
+      "step": 8940
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021363636363636363,
+      "loss": 0.2637,
+      "step": 8941
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0002136239103362391,
+      "loss": 0.1414,
+      "step": 8942
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021361145703611457,
+      "loss": 0.1613,
+      "step": 8943
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021359900373599,
+      "loss": 0.197,
+      "step": 8944
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0002135865504358655,
+      "loss": 0.3065,
+      "step": 8945
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021357409713574095,
+      "loss": 0.3702,
+      "step": 8946
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021356164383561644,
+      "loss": 0.2587,
+      "step": 8947
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0002135491905354919,
+      "loss": 0.2092,
+      "step": 8948
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021353673723536733,
+      "loss": 0.2454,
+      "step": 8949
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021352428393524281,
+      "loss": 0.2165,
+      "step": 8950
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021351183063511827,
+      "loss": 0.34,
+      "step": 8951
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021349937733499376,
+      "loss": 0.3577,
+      "step": 8952
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021348692403486922,
+      "loss": 0.2145,
+      "step": 8953
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.0002134744707347447,
+      "loss": 0.1981,
+      "step": 8954
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021346201743462014,
+      "loss": 0.3817,
+      "step": 8955
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021344956413449562,
+      "loss": 0.1888,
+      "step": 8956
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021343711083437108,
+      "loss": 0.2331,
+      "step": 8957
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021342465753424657,
+      "loss": 0.2986,
+      "step": 8958
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021341220423412203,
+      "loss": 0.3179,
+      "step": 8959
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021339975093399751,
+      "loss": 0.1933,
+      "step": 8960
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00021338729763387295,
+      "loss": 0.2583,
+      "step": 8961
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021337484433374843,
+      "loss": 0.145,
+      "step": 8962
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0002133623910336239,
+      "loss": 0.2156,
+      "step": 8963
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021334993773349935,
+      "loss": 0.2242,
+      "step": 8964
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021333748443337484,
+      "loss": 0.4521,
+      "step": 8965
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021332503113325027,
+      "loss": 0.2929,
+      "step": 8966
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021331257783312576,
+      "loss": 0.2065,
+      "step": 8967
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021330012453300121,
+      "loss": 0.1921,
+      "step": 8968
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0002132876712328767,
+      "loss": 0.2666,
+      "step": 8969
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021327521793275216,
+      "loss": 0.1961,
+      "step": 8970
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021326276463262765,
+      "loss": 0.268,
+      "step": 8971
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021325031133250308,
+      "loss": 0.2112,
+      "step": 8972
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021323785803237856,
+      "loss": 0.1876,
+      "step": 8973
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021322540473225402,
+      "loss": 0.1773,
+      "step": 8974
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0002132129514321295,
+      "loss": 0.1637,
+      "step": 8975
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021320049813200497,
+      "loss": 0.2798,
+      "step": 8976
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021318804483188046,
+      "loss": 0.347,
+      "step": 8977
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0002131755915317559,
+      "loss": 0.2191,
+      "step": 8978
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021316313823163135,
+      "loss": 0.2693,
+      "step": 8979
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021315068493150683,
+      "loss": 0.2011,
+      "step": 8980
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0002131382316313823,
+      "loss": 0.3099,
+      "step": 8981
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021312577833125778,
+      "loss": 0.2044,
+      "step": 8982
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0002131133250311332,
+      "loss": 0.2921,
+      "step": 8983
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0002131008717310087,
+      "loss": 0.3208,
+      "step": 8984
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021308841843088416,
+      "loss": 0.3527,
+      "step": 8985
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00021307596513075964,
+      "loss": 0.2446,
+      "step": 8986
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0002130635118306351,
+      "loss": 0.229,
+      "step": 8987
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0002130510585305106,
+      "loss": 0.3087,
+      "step": 8988
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021303860523038602,
+      "loss": 0.2591,
+      "step": 8989
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0002130261519302615,
+      "loss": 0.4133,
+      "step": 8990
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021301369863013697,
+      "loss": 0.2007,
+      "step": 8991
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021300124533001245,
+      "loss": 0.1748,
+      "step": 8992
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0002129887920298879,
+      "loss": 0.2065,
+      "step": 8993
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0002129763387297634,
+      "loss": 0.1354,
+      "step": 8994
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021296388542963883,
+      "loss": 0.2261,
+      "step": 8995
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0002129514321295143,
+      "loss": 0.3168,
+      "step": 8996
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021293897882938977,
+      "loss": 0.2093,
+      "step": 8997
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021292652552926523,
+      "loss": 0.2262,
+      "step": 8998
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021291407222914072,
+      "loss": 0.186,
+      "step": 8999
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021290161892901615,
+      "loss": 0.1785,
+      "step": 9000
+    },
+    {
+      "epoch": 3.45,
+      "eval_cer": 0.10722437910919484,
+      "eval_loss": 0.38199561834335327,
+      "eval_runtime": 46.7866,
+      "eval_samples_per_second": 37.468,
+      "eval_steps_per_second": 4.702,
+      "eval_wer": 0.17963977010453908,
+      "step": 9000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021288916562889164,
+      "loss": 0.2671,
+      "step": 9001
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0002128767123287671,
+      "loss": 0.2799,
+      "step": 9002
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021286425902864258,
+      "loss": 0.3613,
+      "step": 9003
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021285180572851804,
+      "loss": 0.2448,
+      "step": 9004
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021283935242839353,
+      "loss": 0.2572,
+      "step": 9005
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021282689912826896,
+      "loss": 0.2791,
+      "step": 9006
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021281444582814445,
+      "loss": 0.3357,
+      "step": 9007
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0002128019925280199,
+      "loss": 0.2621,
+      "step": 9008
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0002127895392278954,
+      "loss": 0.3891,
+      "step": 9009
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021277708592777085,
+      "loss": 0.2527,
+      "step": 9010
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021276463262764628,
+      "loss": 0.2245,
+      "step": 9011
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021275217932752177,
+      "loss": 0.2921,
+      "step": 9012
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021273972602739723,
+      "loss": 0.2124,
+      "step": 9013
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00021272727272727272,
+      "loss": 0.1973,
+      "step": 9014
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021271481942714818,
+      "loss": 0.3345,
+      "step": 9015
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021270236612702366,
+      "loss": 0.2435,
+      "step": 9016
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0002126899128268991,
+      "loss": 0.1992,
+      "step": 9017
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021267745952677458,
+      "loss": 0.276,
+      "step": 9018
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021266500622665004,
+      "loss": 0.1888,
+      "step": 9019
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021265255292652553,
+      "loss": 0.203,
+      "step": 9020
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021264009962640098,
+      "loss": 0.3712,
+      "step": 9021
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021262764632627647,
+      "loss": 0.2505,
+      "step": 9022
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0002126151930261519,
+      "loss": 0.5655,
+      "step": 9023
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0002126027397260274,
+      "loss": 0.2128,
+      "step": 9024
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021259028642590285,
+      "loss": 0.1851,
+      "step": 9025
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0002125778331257783,
+      "loss": 0.2119,
+      "step": 9026
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0002125653798256538,
+      "loss": 0.5295,
+      "step": 9027
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021255292652552923,
+      "loss": 0.2354,
+      "step": 9028
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0002125404732254047,
+      "loss": 0.2111,
+      "step": 9029
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021252801992528017,
+      "loss": 0.218,
+      "step": 9030
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021251556662515566,
+      "loss": 0.2169,
+      "step": 9031
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021250311332503112,
+      "loss": 0.2249,
+      "step": 9032
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0002124906600249066,
+      "loss": 0.2506,
+      "step": 9033
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021247820672478203,
+      "loss": 0.3042,
+      "step": 9034
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021246575342465752,
+      "loss": 0.2659,
+      "step": 9035
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021245330012453298,
+      "loss": 0.4452,
+      "step": 9036
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021244084682440847,
+      "loss": 0.2654,
+      "step": 9037
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021242839352428393,
+      "loss": 0.1639,
+      "step": 9038
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021241594022415938,
+      "loss": 0.2138,
+      "step": 9039
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00021240348692403484,
+      "loss": 0.4626,
+      "step": 9040
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021239103362391033,
+      "loss": 0.2117,
+      "step": 9041
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0002123785803237858,
+      "loss": 0.2619,
+      "step": 9042
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021236612702366125,
+      "loss": 0.2131,
+      "step": 9043
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0002123536737235367,
+      "loss": 0.2126,
+      "step": 9044
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021234122042341217,
+      "loss": 0.3321,
+      "step": 9045
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021232876712328765,
+      "loss": 0.6312,
+      "step": 9046
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0002123163138231631,
+      "loss": 0.2765,
+      "step": 9047
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0002123038605230386,
+      "loss": 0.2276,
+      "step": 9048
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021229140722291406,
+      "loss": 0.2467,
+      "step": 9049
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021227895392278952,
+      "loss": 0.1813,
+      "step": 9050
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021226650062266498,
+      "loss": 0.2489,
+      "step": 9051
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021225404732254046,
+      "loss": 0.268,
+      "step": 9052
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021224159402241592,
+      "loss": 0.3005,
+      "step": 9053
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0002122291407222914,
+      "loss": 0.2832,
+      "step": 9054
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021221668742216684,
+      "loss": 0.1984,
+      "step": 9055
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021220423412204233,
+      "loss": 0.1532,
+      "step": 9056
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021219178082191779,
+      "loss": 0.2024,
+      "step": 9057
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021217932752179324,
+      "loss": 0.2074,
+      "step": 9058
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021216687422166873,
+      "loss": 0.5424,
+      "step": 9059
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0002121544209215442,
+      "loss": 0.1853,
+      "step": 9060
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021214196762141965,
+      "loss": 0.2789,
+      "step": 9061
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0002121295143212951,
+      "loss": 0.2238,
+      "step": 9062
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0002121170610211706,
+      "loss": 0.177,
+      "step": 9063
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021210460772104605,
+      "loss": 0.2985,
+      "step": 9064
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021209215442092154,
+      "loss": 0.3248,
+      "step": 9065
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00021207970112079697,
+      "loss": 0.2121,
+      "step": 9066
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021206724782067246,
+      "loss": 0.1982,
+      "step": 9067
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021205479452054792,
+      "loss": 0.1506,
+      "step": 9068
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0002120423412204234,
+      "loss": 0.1985,
+      "step": 9069
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021202988792029886,
+      "loss": 0.2871,
+      "step": 9070
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021201743462017435,
+      "loss": 0.2877,
+      "step": 9071
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021200498132004978,
+      "loss": 0.2238,
+      "step": 9072
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021199252801992524,
+      "loss": 0.2997,
+      "step": 9073
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021198007471980073,
+      "loss": 0.2861,
+      "step": 9074
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021196762141967619,
+      "loss": 0.175,
+      "step": 9075
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021195516811955167,
+      "loss": 0.3036,
+      "step": 9076
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0002119427148194271,
+      "loss": 0.4249,
+      "step": 9077
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0002119302615193026,
+      "loss": 0.3991,
+      "step": 9078
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021191780821917805,
+      "loss": 0.198,
+      "step": 9079
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021190535491905354,
+      "loss": 0.315,
+      "step": 9080
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.000211892901618929,
+      "loss": 0.1742,
+      "step": 9081
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021188044831880448,
+      "loss": 0.1629,
+      "step": 9082
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021186799501867991,
+      "loss": 0.2643,
+      "step": 9083
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0002118555417185554,
+      "loss": 0.3062,
+      "step": 9084
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021184308841843086,
+      "loss": 0.2791,
+      "step": 9085
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021183063511830635,
+      "loss": 0.2015,
+      "step": 9086
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0002118181818181818,
+      "loss": 0.2336,
+      "step": 9087
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0002118057285180573,
+      "loss": 0.1828,
+      "step": 9088
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021179327521793272,
+      "loss": 0.2205,
+      "step": 9089
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021178082191780818,
+      "loss": 0.3699,
+      "step": 9090
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021176836861768367,
+      "loss": 0.2406,
+      "step": 9091
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00021175591531755913,
+      "loss": 0.144,
+      "step": 9092
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021174346201743461,
+      "loss": 0.2932,
+      "step": 9093
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021173100871731005,
+      "loss": 0.2938,
+      "step": 9094
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021171855541718553,
+      "loss": 0.1313,
+      "step": 9095
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000211706102117061,
+      "loss": 0.348,
+      "step": 9096
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021169364881693648,
+      "loss": 0.2996,
+      "step": 9097
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021168119551681194,
+      "loss": 0.1637,
+      "step": 9098
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021166874221668742,
+      "loss": 0.2024,
+      "step": 9099
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021165628891656286,
+      "loss": 0.3204,
+      "step": 9100
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021164383561643834,
+      "loss": 0.3042,
+      "step": 9101
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0002116313823163138,
+      "loss": 0.5826,
+      "step": 9102
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0002116189290161893,
+      "loss": 0.2968,
+      "step": 9103
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021160647571606475,
+      "loss": 0.1545,
+      "step": 9104
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021159402241594018,
+      "loss": 0.2516,
+      "step": 9105
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021158156911581566,
+      "loss": 0.2085,
+      "step": 9106
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021156911581569112,
+      "loss": 0.1676,
+      "step": 9107
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0002115566625155666,
+      "loss": 0.2682,
+      "step": 9108
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021154420921544207,
+      "loss": 0.3776,
+      "step": 9109
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021153175591531756,
+      "loss": 0.2192,
+      "step": 9110
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000211519302615193,
+      "loss": 0.2764,
+      "step": 9111
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021150684931506847,
+      "loss": 0.1772,
+      "step": 9112
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021149439601494393,
+      "loss": 0.1802,
+      "step": 9113
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021148194271481942,
+      "loss": 0.1666,
+      "step": 9114
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021146948941469488,
+      "loss": 0.3974,
+      "step": 9115
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021145703611457036,
+      "loss": 0.188,
+      "step": 9116
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0002114445828144458,
+      "loss": 0.2376,
+      "step": 9117
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00021143212951432128,
+      "loss": 0.1949,
+      "step": 9118
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021141967621419674,
+      "loss": 0.2113,
+      "step": 9119
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0002114072229140722,
+      "loss": 0.17,
+      "step": 9120
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0002113947696139477,
+      "loss": 0.286,
+      "step": 9121
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021138231631382312,
+      "loss": 0.2407,
+      "step": 9122
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0002113698630136986,
+      "loss": 0.2244,
+      "step": 9123
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021135740971357406,
+      "loss": 0.1996,
+      "step": 9124
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021134495641344955,
+      "loss": 0.1845,
+      "step": 9125
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.000211332503113325,
+      "loss": 0.1699,
+      "step": 9126
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0002113200498132005,
+      "loss": 0.3347,
+      "step": 9127
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021130759651307593,
+      "loss": 0.3972,
+      "step": 9128
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021129514321295141,
+      "loss": 0.3042,
+      "step": 9129
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021128268991282687,
+      "loss": 0.1927,
+      "step": 9130
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021127023661270236,
+      "loss": 0.2343,
+      "step": 9131
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021125778331257782,
+      "loss": 0.2598,
+      "step": 9132
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0002112453300124533,
+      "loss": 0.1375,
+      "step": 9133
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021123287671232874,
+      "loss": 0.2975,
+      "step": 9134
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021122042341220422,
+      "loss": 0.3171,
+      "step": 9135
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021120797011207968,
+      "loss": 0.3114,
+      "step": 9136
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021119551681195514,
+      "loss": 0.2493,
+      "step": 9137
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021118306351183063,
+      "loss": 0.1471,
+      "step": 9138
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021117061021170606,
+      "loss": 0.2096,
+      "step": 9139
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021115815691158155,
+      "loss": 0.3194,
+      "step": 9140
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.000211145703611457,
+      "loss": 0.2181,
+      "step": 9141
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0002111332503113325,
+      "loss": 0.3101,
+      "step": 9142
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021112079701120795,
+      "loss": 0.2497,
+      "step": 9143
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00021110834371108344,
+      "loss": 0.2474,
+      "step": 9144
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021109589041095887,
+      "loss": 0.2451,
+      "step": 9145
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021108343711083436,
+      "loss": 0.3895,
+      "step": 9146
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021107098381070982,
+      "loss": 0.2237,
+      "step": 9147
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0002110585305105853,
+      "loss": 0.2437,
+      "step": 9148
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021104607721046076,
+      "loss": 0.2139,
+      "step": 9149
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021103362391033625,
+      "loss": 0.2191,
+      "step": 9150
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021102117061021168,
+      "loss": 0.2323,
+      "step": 9151
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021100871731008714,
+      "loss": 0.4969,
+      "step": 9152
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021099626400996262,
+      "loss": 0.2883,
+      "step": 9153
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021098381070983808,
+      "loss": 0.2451,
+      "step": 9154
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021097135740971357,
+      "loss": 0.2576,
+      "step": 9155
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.000210958904109589,
+      "loss": 0.1405,
+      "step": 9156
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0002109464508094645,
+      "loss": 0.2122,
+      "step": 9157
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021093399750933995,
+      "loss": 0.2105,
+      "step": 9158
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021092154420921543,
+      "loss": 0.3656,
+      "step": 9159
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0002109090909090909,
+      "loss": 0.2084,
+      "step": 9160
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021089663760896638,
+      "loss": 0.2552,
+      "step": 9161
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0002108841843088418,
+      "loss": 0.2938,
+      "step": 9162
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0002108717310087173,
+      "loss": 0.2079,
+      "step": 9163
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021085927770859276,
+      "loss": 0.2343,
+      "step": 9164
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021084682440846824,
+      "loss": 0.3017,
+      "step": 9165
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0002108343711083437,
+      "loss": 0.2315,
+      "step": 9166
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0002108219178082192,
+      "loss": 0.1502,
+      "step": 9167
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021080946450809462,
+      "loss": 0.1458,
+      "step": 9168
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021079701120797008,
+      "loss": 0.3622,
+      "step": 9169
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00021078455790784557,
+      "loss": 0.3178,
+      "step": 9170
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021077210460772103,
+      "loss": 0.347,
+      "step": 9171
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0002107596513075965,
+      "loss": 0.3382,
+      "step": 9172
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021074719800747194,
+      "loss": 0.2921,
+      "step": 9173
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021073474470734743,
+      "loss": 0.2042,
+      "step": 9174
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0002107222914072229,
+      "loss": 0.1796,
+      "step": 9175
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021070983810709838,
+      "loss": 0.2109,
+      "step": 9176
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021069738480697383,
+      "loss": 0.4191,
+      "step": 9177
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021068493150684932,
+      "loss": 0.2031,
+      "step": 9178
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021067247820672475,
+      "loss": 0.2525,
+      "step": 9179
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021066002490660024,
+      "loss": 0.2317,
+      "step": 9180
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0002106475716064757,
+      "loss": 0.1927,
+      "step": 9181
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021063511830635118,
+      "loss": 0.1833,
+      "step": 9182
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021062266500622664,
+      "loss": 0.4422,
+      "step": 9183
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021061021170610208,
+      "loss": 0.3743,
+      "step": 9184
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021059775840597756,
+      "loss": 0.2116,
+      "step": 9185
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021058530510585302,
+      "loss": 0.3013,
+      "step": 9186
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0002105728518057285,
+      "loss": 0.2635,
+      "step": 9187
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021056039850560397,
+      "loss": 0.2976,
+      "step": 9188
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021054794520547945,
+      "loss": 0.2677,
+      "step": 9189
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021053549190535488,
+      "loss": 0.5193,
+      "step": 9190
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021052303860523037,
+      "loss": 0.1913,
+      "step": 9191
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021051058530510583,
+      "loss": 0.2776,
+      "step": 9192
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021049813200498132,
+      "loss": 0.2902,
+      "step": 9193
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021048567870485678,
+      "loss": 0.1364,
+      "step": 9194
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00021047322540473226,
+      "loss": 0.2886,
+      "step": 9195
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0002104607721046077,
+      "loss": 0.4187,
+      "step": 9196
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021044831880448318,
+      "loss": 0.3062,
+      "step": 9197
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021043586550435864,
+      "loss": 0.2077,
+      "step": 9198
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0002104234122042341,
+      "loss": 0.3503,
+      "step": 9199
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021041095890410959,
+      "loss": 0.2961,
+      "step": 9200
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021039850560398502,
+      "loss": 0.2148,
+      "step": 9201
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0002103860523038605,
+      "loss": 0.4274,
+      "step": 9202
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021037359900373596,
+      "loss": 0.3127,
+      "step": 9203
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021036114570361145,
+      "loss": 0.2721,
+      "step": 9204
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0002103486924034869,
+      "loss": 0.1915,
+      "step": 9205
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021033623910336237,
+      "loss": 0.1465,
+      "step": 9206
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021032378580323783,
+      "loss": 0.1849,
+      "step": 9207
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0002103113325031133,
+      "loss": 0.287,
+      "step": 9208
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021029887920298877,
+      "loss": 0.4202,
+      "step": 9209
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021028642590286426,
+      "loss": 0.2868,
+      "step": 9210
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021027397260273972,
+      "loss": 0.22,
+      "step": 9211
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021026151930261518,
+      "loss": 0.1599,
+      "step": 9212
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021024906600249064,
+      "loss": 0.2862,
+      "step": 9213
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021023661270236612,
+      "loss": 0.3379,
+      "step": 9214
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021022415940224158,
+      "loss": 0.5771,
+      "step": 9215
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021021170610211704,
+      "loss": 0.2383,
+      "step": 9216
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0002101992528019925,
+      "loss": 0.2213,
+      "step": 9217
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021018679950186796,
+      "loss": 0.3362,
+      "step": 9218
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021017434620174344,
+      "loss": 0.1913,
+      "step": 9219
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0002101618929016189,
+      "loss": 0.418,
+      "step": 9220
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0002101494396014944,
+      "loss": 0.3085,
+      "step": 9221
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00021013698630136985,
+      "loss": 0.3,
+      "step": 9222
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0002101245330012453,
+      "loss": 0.3014,
+      "step": 9223
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00021011207970112077,
+      "loss": 0.2808,
+      "step": 9224
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00021009962640099625,
+      "loss": 0.2357,
+      "step": 9225
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0002100871731008717,
+      "loss": 0.2911,
+      "step": 9226
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0002100747198007472,
+      "loss": 0.4734,
+      "step": 9227
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00021006226650062263,
+      "loss": 0.2443,
+      "step": 9228
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00021004981320049812,
+      "loss": 0.2053,
+      "step": 9229
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00021003735990037358,
+      "loss": 0.2675,
+      "step": 9230
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00021002490660024904,
+      "loss": 0.1339,
+      "step": 9231
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00021001245330012452,
+      "loss": 0.2458,
+      "step": 9232
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 0.2759,
+      "step": 9233
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00020998754669987544,
+      "loss": 0.2853,
+      "step": 9234
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0002099750933997509,
+      "loss": 0.3177,
+      "step": 9235
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00020996264009962639,
+      "loss": 0.1938,
+      "step": 9236
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00020995018679950185,
+      "loss": 0.2616,
+      "step": 9237
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00020993773349937733,
+      "loss": 0.1982,
+      "step": 9238
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00020992528019925276,
+      "loss": 0.1855,
+      "step": 9239
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00020991282689912825,
+      "loss": 0.2842,
+      "step": 9240
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0002099003735990037,
+      "loss": 0.2415,
+      "step": 9241
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0002098879202988792,
+      "loss": 0.2624,
+      "step": 9242
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00020987546699875465,
+      "loss": 0.2775,
+      "step": 9243
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00020986301369863014,
+      "loss": 0.2369,
+      "step": 9244
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00020985056039850557,
+      "loss": 0.188,
+      "step": 9245
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00020983810709838103,
+      "loss": 0.4099,
+      "step": 9246
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00020982565379825652,
+      "loss": 0.2169,
+      "step": 9247
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00020981320049813198,
+      "loss": 0.2844,
+      "step": 9248
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020980074719800746,
+      "loss": 0.2651,
+      "step": 9249
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0002097882938978829,
+      "loss": 0.2902,
+      "step": 9250
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020977584059775838,
+      "loss": 0.2308,
+      "step": 9251
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020976338729763384,
+      "loss": 0.483,
+      "step": 9252
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020975093399750933,
+      "loss": 0.1955,
+      "step": 9253
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0002097384806973848,
+      "loss": 0.2798,
+      "step": 9254
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020972602739726027,
+      "loss": 0.1713,
+      "step": 9255
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0002097135740971357,
+      "loss": 0.1763,
+      "step": 9256
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0002097011207970112,
+      "loss": 0.2467,
+      "step": 9257
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020968866749688665,
+      "loss": 0.2961,
+      "step": 9258
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020967621419676214,
+      "loss": 0.3001,
+      "step": 9259
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0002096637608966376,
+      "loss": 0.3652,
+      "step": 9260
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020965130759651308,
+      "loss": 0.2457,
+      "step": 9261
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020963885429638851,
+      "loss": 0.1544,
+      "step": 9262
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020962640099626397,
+      "loss": 0.218,
+      "step": 9263
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020961394769613946,
+      "loss": 0.2804,
+      "step": 9264
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020960149439601492,
+      "loss": 0.3047,
+      "step": 9265
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0002095890410958904,
+      "loss": 0.2361,
+      "step": 9266
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020957658779576584,
+      "loss": 0.213,
+      "step": 9267
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020956413449564132,
+      "loss": 0.319,
+      "step": 9268
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020955168119551678,
+      "loss": 0.2214,
+      "step": 9269
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020953922789539227,
+      "loss": 0.1925,
+      "step": 9270
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020952677459526773,
+      "loss": 0.3593,
+      "step": 9271
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020951432129514321,
+      "loss": 0.269,
+      "step": 9272
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020950186799501865,
+      "loss": 0.2126,
+      "step": 9273
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00020948941469489413,
+      "loss": 0.1884,
+      "step": 9274
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002094769613947696,
+      "loss": 0.3445,
+      "step": 9275
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020946450809464508,
+      "loss": 0.1492,
+      "step": 9276
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020945205479452054,
+      "loss": 0.4666,
+      "step": 9277
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020943960149439597,
+      "loss": 0.293,
+      "step": 9278
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020942714819427146,
+      "loss": 0.2537,
+      "step": 9279
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020941469489414691,
+      "loss": 0.1805,
+      "step": 9280
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002094022415940224,
+      "loss": 0.184,
+      "step": 9281
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020938978829389786,
+      "loss": 0.2137,
+      "step": 9282
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020937733499377335,
+      "loss": 0.2689,
+      "step": 9283
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020936488169364878,
+      "loss": 0.42,
+      "step": 9284
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020935242839352426,
+      "loss": 0.3137,
+      "step": 9285
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020933997509339972,
+      "loss": 0.2876,
+      "step": 9286
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002093275217932752,
+      "loss": 0.2016,
+      "step": 9287
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020931506849315067,
+      "loss": 0.2169,
+      "step": 9288
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020930261519302616,
+      "loss": 0.256,
+      "step": 9289
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002092901618929016,
+      "loss": 0.3845,
+      "step": 9290
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020927770859277707,
+      "loss": 0.2517,
+      "step": 9291
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020926525529265253,
+      "loss": 0.2854,
+      "step": 9292
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.000209252801992528,
+      "loss": 0.2403,
+      "step": 9293
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020924034869240348,
+      "loss": 0.1578,
+      "step": 9294
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002092278953922789,
+      "loss": 0.2177,
+      "step": 9295
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002092154420921544,
+      "loss": 0.492,
+      "step": 9296
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020920298879202986,
+      "loss": 0.304,
+      "step": 9297
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020919053549190534,
+      "loss": 0.483,
+      "step": 9298
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002091780821917808,
+      "loss": 0.2738,
+      "step": 9299
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002091656288916563,
+      "loss": 0.344,
+      "step": 9300
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00020915317559153172,
+      "loss": 0.2736,
+      "step": 9301
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0002091407222914072,
+      "loss": 0.4412,
+      "step": 9302
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020912826899128267,
+      "loss": 0.2895,
+      "step": 9303
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020911581569115815,
+      "loss": 0.2472,
+      "step": 9304
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0002091033623910336,
+      "loss": 0.2159,
+      "step": 9305
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0002090909090909091,
+      "loss": 0.2712,
+      "step": 9306
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020907845579078453,
+      "loss": 0.1601,
+      "step": 9307
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020906600249066002,
+      "loss": 0.3323,
+      "step": 9308
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020905354919053547,
+      "loss": 0.4573,
+      "step": 9309
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020904109589041093,
+      "loss": 0.1991,
+      "step": 9310
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020902864259028642,
+      "loss": 0.3951,
+      "step": 9311
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020901618929016185,
+      "loss": 0.2238,
+      "step": 9312
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020900373599003734,
+      "loss": 0.2989,
+      "step": 9313
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0002089912826899128,
+      "loss": 0.1957,
+      "step": 9314
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020897882938978828,
+      "loss": 0.3384,
+      "step": 9315
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020896637608966374,
+      "loss": 0.31,
+      "step": 9316
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020895392278953923,
+      "loss": 0.2036,
+      "step": 9317
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020894146948941466,
+      "loss": 0.2492,
+      "step": 9318
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020892901618929015,
+      "loss": 0.1981,
+      "step": 9319
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0002089165628891656,
+      "loss": 0.2845,
+      "step": 9320
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0002089041095890411,
+      "loss": 0.4034,
+      "step": 9321
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020889165628891655,
+      "loss": 0.2311,
+      "step": 9322
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020887920298879204,
+      "loss": 0.2067,
+      "step": 9323
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020886674968866747,
+      "loss": 0.2013,
+      "step": 9324
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020885429638854293,
+      "loss": 0.1616,
+      "step": 9325
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020884184308841842,
+      "loss": 0.1894,
+      "step": 9326
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00020882938978829388,
+      "loss": 0.3077,
+      "step": 9327
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020881693648816936,
+      "loss": 0.2207,
+      "step": 9328
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0002088044831880448,
+      "loss": 0.2462,
+      "step": 9329
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020879202988792028,
+      "loss": 0.2265,
+      "step": 9330
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020877957658779574,
+      "loss": 0.1951,
+      "step": 9331
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020876712328767123,
+      "loss": 0.19,
+      "step": 9332
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020875466998754668,
+      "loss": 0.2214,
+      "step": 9333
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020874221668742217,
+      "loss": 0.4899,
+      "step": 9334
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0002087297633872976,
+      "loss": 0.3603,
+      "step": 9335
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0002087173100871731,
+      "loss": 0.2306,
+      "step": 9336
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020870485678704855,
+      "loss": 0.3113,
+      "step": 9337
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020869240348692403,
+      "loss": 0.2665,
+      "step": 9338
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0002086799501867995,
+      "loss": 0.2505,
+      "step": 9339
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020866749688667493,
+      "loss": 0.4692,
+      "step": 9340
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0002086550435865504,
+      "loss": 0.2884,
+      "step": 9341
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020864259028642587,
+      "loss": 0.1838,
+      "step": 9342
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020863013698630136,
+      "loss": 0.2732,
+      "step": 9343
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020861768368617682,
+      "loss": 0.1642,
+      "step": 9344
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0002086052303860523,
+      "loss": 0.2772,
+      "step": 9345
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020859277708592774,
+      "loss": 0.357,
+      "step": 9346
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020858032378580322,
+      "loss": 0.202,
+      "step": 9347
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020856787048567868,
+      "loss": 0.2215,
+      "step": 9348
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020855541718555417,
+      "loss": 0.4498,
+      "step": 9349
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020854296388542963,
+      "loss": 0.4773,
+      "step": 9350
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0002085305105853051,
+      "loss": 0.199,
+      "step": 9351
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020851805728518054,
+      "loss": 0.2749,
+      "step": 9352
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00020850560398505603,
+      "loss": 0.2361,
+      "step": 9353
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0002084931506849315,
+      "loss": 0.1817,
+      "step": 9354
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020848069738480698,
+      "loss": 0.3396,
+      "step": 9355
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020846824408468244,
+      "loss": 0.2213,
+      "step": 9356
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020845579078455787,
+      "loss": 0.2713,
+      "step": 9357
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020844333748443335,
+      "loss": 0.3503,
+      "step": 9358
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0002084308841843088,
+      "loss": 0.3127,
+      "step": 9359
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0002084184308841843,
+      "loss": 0.2994,
+      "step": 9360
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020840597758405976,
+      "loss": 0.1998,
+      "step": 9361
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020839352428393524,
+      "loss": 0.1963,
+      "step": 9362
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020838107098381068,
+      "loss": 0.2406,
+      "step": 9363
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020836861768368616,
+      "loss": 0.3238,
+      "step": 9364
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020835616438356162,
+      "loss": 0.4366,
+      "step": 9365
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0002083437110834371,
+      "loss": 0.2184,
+      "step": 9366
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020833125778331257,
+      "loss": 0.2429,
+      "step": 9367
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020831880448318803,
+      "loss": 0.2522,
+      "step": 9368
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020830635118306349,
+      "loss": 0.1998,
+      "step": 9369
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020829389788293897,
+      "loss": 0.2627,
+      "step": 9370
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020828144458281443,
+      "loss": 0.3218,
+      "step": 9371
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0002082689912826899,
+      "loss": 0.3115,
+      "step": 9372
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020825653798256538,
+      "loss": 0.2267,
+      "step": 9373
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0002082440846824408,
+      "loss": 0.1622,
+      "step": 9374
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0002082316313823163,
+      "loss": 0.2298,
+      "step": 9375
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020821917808219175,
+      "loss": 0.2922,
+      "step": 9376
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020820672478206724,
+      "loss": 0.3426,
+      "step": 9377
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0002081942714819427,
+      "loss": 0.3094,
+      "step": 9378
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00020818181818181816,
+      "loss": 0.2253,
+      "step": 9379
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020816936488169362,
+      "loss": 0.2685,
+      "step": 9380
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0002081569115815691,
+      "loss": 0.1744,
+      "step": 9381
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020814445828144456,
+      "loss": 0.345,
+      "step": 9382
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020813200498132005,
+      "loss": 0.4053,
+      "step": 9383
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0002081195516811955,
+      "loss": 0.4431,
+      "step": 9384
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020810709838107097,
+      "loss": 0.2298,
+      "step": 9385
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020809464508094643,
+      "loss": 0.2583,
+      "step": 9386
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020808219178082189,
+      "loss": 0.1655,
+      "step": 9387
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020806973848069737,
+      "loss": 0.224,
+      "step": 9388
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020805728518057283,
+      "loss": 0.3121,
+      "step": 9389
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0002080448318804483,
+      "loss": 0.4467,
+      "step": 9390
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020803237858032375,
+      "loss": 0.2512,
+      "step": 9391
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020801992528019924,
+      "loss": 0.2307,
+      "step": 9392
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0002080074719800747,
+      "loss": 0.2093,
+      "step": 9393
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020799501867995018,
+      "loss": 0.191,
+      "step": 9394
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020798256537982564,
+      "loss": 0.2324,
+      "step": 9395
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0002079701120797011,
+      "loss": 0.3,
+      "step": 9396
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020795765877957656,
+      "loss": 0.4627,
+      "step": 9397
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020794520547945205,
+      "loss": 0.2031,
+      "step": 9398
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0002079327521793275,
+      "loss": 0.2677,
+      "step": 9399
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.000207920298879203,
+      "loss": 0.2287,
+      "step": 9400
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020790784557907842,
+      "loss": 0.207,
+      "step": 9401
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0002078953922789539,
+      "loss": 0.2333,
+      "step": 9402
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020788293897882937,
+      "loss": 0.2463,
+      "step": 9403
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020787048567870483,
+      "loss": 0.1512,
+      "step": 9404
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00020785803237858031,
+      "loss": 0.2402,
+      "step": 9405
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020784557907845577,
+      "loss": 0.2044,
+      "step": 9406
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020783312577833123,
+      "loss": 0.2805,
+      "step": 9407
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0002078206724782067,
+      "loss": 0.2409,
+      "step": 9408
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020780821917808218,
+      "loss": 0.3728,
+      "step": 9409
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020779576587795764,
+      "loss": 0.2065,
+      "step": 9410
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020778331257783312,
+      "loss": 0.2185,
+      "step": 9411
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020777085927770856,
+      "loss": 0.1815,
+      "step": 9412
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020775840597758404,
+      "loss": 0.2646,
+      "step": 9413
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0002077459526774595,
+      "loss": 0.1856,
+      "step": 9414
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.000207733499377335,
+      "loss": 0.3321,
+      "step": 9415
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020772104607721045,
+      "loss": 0.3634,
+      "step": 9416
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020770859277708593,
+      "loss": 0.2241,
+      "step": 9417
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020769613947696136,
+      "loss": 0.1558,
+      "step": 9418
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020768368617683682,
+      "loss": 0.1836,
+      "step": 9419
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0002076712328767123,
+      "loss": 0.2858,
+      "step": 9420
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020765877957658777,
+      "loss": 0.2227,
+      "step": 9421
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020764632627646326,
+      "loss": 0.1782,
+      "step": 9422
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0002076338729763387,
+      "loss": 0.2001,
+      "step": 9423
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020762141967621417,
+      "loss": 0.2232,
+      "step": 9424
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020760896637608963,
+      "loss": 0.1553,
+      "step": 9425
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020759651307596512,
+      "loss": 0.3771,
+      "step": 9426
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020758405977584058,
+      "loss": 0.4935,
+      "step": 9427
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020757160647571606,
+      "loss": 0.2022,
+      "step": 9428
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0002075591531755915,
+      "loss": 0.2062,
+      "step": 9429
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020754669987546698,
+      "loss": 0.2081,
+      "step": 9430
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00020753424657534244,
+      "loss": 0.2602,
+      "step": 9431
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020752179327521793,
+      "loss": 0.2426,
+      "step": 9432
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0002075093399750934,
+      "loss": 0.305,
+      "step": 9433
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020749688667496887,
+      "loss": 0.3016,
+      "step": 9434
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0002074844333748443,
+      "loss": 0.3835,
+      "step": 9435
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020747198007471976,
+      "loss": 0.201,
+      "step": 9436
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020745952677459525,
+      "loss": 0.1981,
+      "step": 9437
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0002074470734744707,
+      "loss": 0.284,
+      "step": 9438
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0002074346201743462,
+      "loss": 0.2811,
+      "step": 9439
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020742216687422163,
+      "loss": 0.5363,
+      "step": 9440
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020740971357409712,
+      "loss": 0.184,
+      "step": 9441
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020739726027397257,
+      "loss": 0.1947,
+      "step": 9442
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020738480697384806,
+      "loss": 0.1081,
+      "step": 9443
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020737235367372352,
+      "loss": 0.134,
+      "step": 9444
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.000207359900373599,
+      "loss": 0.2045,
+      "step": 9445
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020734744707347444,
+      "loss": 0.4111,
+      "step": 9446
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020733499377334992,
+      "loss": 0.1907,
+      "step": 9447
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020732254047322538,
+      "loss": 0.1742,
+      "step": 9448
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020731008717310087,
+      "loss": 0.247,
+      "step": 9449
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020729763387297633,
+      "loss": 0.1811,
+      "step": 9450
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020728518057285176,
+      "loss": 0.2367,
+      "step": 9451
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020727272727272725,
+      "loss": 0.3987,
+      "step": 9452
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0002072602739726027,
+      "loss": 0.3396,
+      "step": 9453
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0002072478206724782,
+      "loss": 0.2099,
+      "step": 9454
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020723536737235365,
+      "loss": 0.1782,
+      "step": 9455
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020722291407222914,
+      "loss": 0.3709,
+      "step": 9456
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00020721046077210457,
+      "loss": 0.1878,
+      "step": 9457
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020719800747198006,
+      "loss": 0.2236,
+      "step": 9458
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020718555417185552,
+      "loss": 0.5668,
+      "step": 9459
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.000207173100871731,
+      "loss": 0.2025,
+      "step": 9460
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020716064757160646,
+      "loss": 0.1759,
+      "step": 9461
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020714819427148195,
+      "loss": 0.1571,
+      "step": 9462
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020713574097135738,
+      "loss": 0.2322,
+      "step": 9463
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020712328767123287,
+      "loss": 0.2625,
+      "step": 9464
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020711083437110832,
+      "loss": 0.4505,
+      "step": 9465
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020709838107098378,
+      "loss": 0.2468,
+      "step": 9466
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020708592777085927,
+      "loss": 0.2094,
+      "step": 9467
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0002070734744707347,
+      "loss": 0.1605,
+      "step": 9468
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0002070610211706102,
+      "loss": 0.1905,
+      "step": 9469
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020704856787048565,
+      "loss": 0.2213,
+      "step": 9470
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020703611457036113,
+      "loss": 0.3172,
+      "step": 9471
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0002070236612702366,
+      "loss": 0.2722,
+      "step": 9472
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020701120797011208,
+      "loss": 0.284,
+      "step": 9473
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0002069987546699875,
+      "loss": 0.1942,
+      "step": 9474
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.000206986301369863,
+      "loss": 0.2261,
+      "step": 9475
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020697384806973846,
+      "loss": 0.2952,
+      "step": 9476
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020696139476961394,
+      "loss": 0.4371,
+      "step": 9477
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0002069489414694894,
+      "loss": 0.3496,
+      "step": 9478
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0002069364881693649,
+      "loss": 0.2545,
+      "step": 9479
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020692403486924032,
+      "loss": 0.1839,
+      "step": 9480
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0002069115815691158,
+      "loss": 0.2453,
+      "step": 9481
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020689912826899127,
+      "loss": 0.1782,
+      "step": 9482
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00020688667496886673,
+      "loss": 0.3433,
+      "step": 9483
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0002068742216687422,
+      "loss": 0.4016,
+      "step": 9484
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020686176836861764,
+      "loss": 0.1998,
+      "step": 9485
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020684931506849313,
+      "loss": 0.249,
+      "step": 9486
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0002068368617683686,
+      "loss": 0.2761,
+      "step": 9487
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020682440846824408,
+      "loss": 0.1923,
+      "step": 9488
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020681195516811953,
+      "loss": 0.2254,
+      "step": 9489
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020679950186799502,
+      "loss": 0.4038,
+      "step": 9490
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020678704856787045,
+      "loss": 0.4416,
+      "step": 9491
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020677459526774594,
+      "loss": 0.2255,
+      "step": 9492
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0002067621419676214,
+      "loss": 0.1853,
+      "step": 9493
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020674968866749688,
+      "loss": 0.2362,
+      "step": 9494
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020673723536737234,
+      "loss": 0.2558,
+      "step": 9495
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020672478206724783,
+      "loss": 0.4843,
+      "step": 9496
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020671232876712326,
+      "loss": 0.2484,
+      "step": 9497
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020669987546699872,
+      "loss": 0.2958,
+      "step": 9498
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0002066874221668742,
+      "loss": 0.3644,
+      "step": 9499
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020667496886674967,
+      "loss": 0.321,
+      "step": 9500
+    },
+    {
+      "epoch": 3.64,
+      "eval_cer": 0.11264802856195251,
+      "eval_loss": 0.3532786965370178,
+      "eval_runtime": 46.1439,
+      "eval_samples_per_second": 37.99,
+      "eval_steps_per_second": 4.768,
+      "eval_wer": 0.19939358283929945,
+      "step": 9500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020666251556662515,
+      "loss": 0.2491,
+      "step": 9501
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020665006226650059,
+      "loss": 0.3838,
+      "step": 9502
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020663760896637607,
+      "loss": 0.2801,
+      "step": 9503
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020662515566625153,
+      "loss": 0.2013,
+      "step": 9504
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020661270236612702,
+      "loss": 0.1896,
+      "step": 9505
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020660024906600248,
+      "loss": 0.2963,
+      "step": 9506
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020658779576587796,
+      "loss": 0.2661,
+      "step": 9507
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0002065753424657534,
+      "loss": 0.285,
+      "step": 9508
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00020656288916562888,
+      "loss": 0.2945,
+      "step": 9509
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020655043586550434,
+      "loss": 0.1558,
+      "step": 9510
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020653798256537983,
+      "loss": 0.2496,
+      "step": 9511
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020652552926525529,
+      "loss": 0.2721,
+      "step": 9512
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020651307596513072,
+      "loss": 0.3898,
+      "step": 9513
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0002065006226650062,
+      "loss": 0.2979,
+      "step": 9514
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020648816936488166,
+      "loss": 0.4153,
+      "step": 9515
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020647571606475715,
+      "loss": 0.2972,
+      "step": 9516
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0002064632627646326,
+      "loss": 0.3368,
+      "step": 9517
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0002064508094645081,
+      "loss": 0.1852,
+      "step": 9518
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020643835616438353,
+      "loss": 0.1668,
+      "step": 9519
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.000206425902864259,
+      "loss": 0.3173,
+      "step": 9520
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020641344956413447,
+      "loss": 0.4122,
+      "step": 9521
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020640099626400996,
+      "loss": 0.2168,
+      "step": 9522
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020638854296388542,
+      "loss": 0.1943,
+      "step": 9523
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0002063760896637609,
+      "loss": 0.1512,
+      "step": 9524
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020636363636363634,
+      "loss": 0.2339,
+      "step": 9525
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020635118306351182,
+      "loss": 0.2762,
+      "step": 9526
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020633872976338728,
+      "loss": 0.3977,
+      "step": 9527
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020632627646326277,
+      "loss": 0.2551,
+      "step": 9528
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020631382316313823,
+      "loss": 0.2092,
+      "step": 9529
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020630136986301366,
+      "loss": 0.171,
+      "step": 9530
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020628891656288914,
+      "loss": 0.2296,
+      "step": 9531
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0002062764632627646,
+      "loss": 0.2574,
+      "step": 9532
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0002062640099626401,
+      "loss": 0.4689,
+      "step": 9533
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020625155666251555,
+      "loss": 0.4414,
+      "step": 9534
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00020623910336239104,
+      "loss": 0.2765,
+      "step": 9535
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020622665006226647,
+      "loss": 0.1607,
+      "step": 9536
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020621419676214195,
+      "loss": 0.1748,
+      "step": 9537
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0002062017434620174,
+      "loss": 0.1794,
+      "step": 9538
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0002061892901618929,
+      "loss": 0.2242,
+      "step": 9539
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020617683686176836,
+      "loss": 0.3093,
+      "step": 9540
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020616438356164382,
+      "loss": 0.2223,
+      "step": 9541
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020615193026151928,
+      "loss": 0.2349,
+      "step": 9542
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020613947696139476,
+      "loss": 0.2391,
+      "step": 9543
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020612702366127022,
+      "loss": 0.1963,
+      "step": 9544
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020611457036114568,
+      "loss": 0.3274,
+      "step": 9545
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020610211706102117,
+      "loss": 0.2794,
+      "step": 9546
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0002060896637608966,
+      "loss": 0.2521,
+      "step": 9547
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020607721046077209,
+      "loss": 0.1482,
+      "step": 9548
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020606475716064755,
+      "loss": 0.1783,
+      "step": 9549
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020605230386052303,
+      "loss": 0.2995,
+      "step": 9550
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0002060398505603985,
+      "loss": 0.2224,
+      "step": 9551
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020602739726027395,
+      "loss": 0.3227,
+      "step": 9552
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0002060149439601494,
+      "loss": 0.2234,
+      "step": 9553
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0002060024906600249,
+      "loss": 0.2004,
+      "step": 9554
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020599003735990035,
+      "loss": 0.2449,
+      "step": 9555
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020597758405977584,
+      "loss": 0.1362,
+      "step": 9556
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0002059651307596513,
+      "loss": 0.2145,
+      "step": 9557
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020595267745952676,
+      "loss": 0.198,
+      "step": 9558
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020594022415940222,
+      "loss": 0.3314,
+      "step": 9559
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020592777085927768,
+      "loss": 0.199,
+      "step": 9560
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00020591531755915316,
+      "loss": 0.1965,
+      "step": 9561
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020590286425902862,
+      "loss": 0.2005,
+      "step": 9562
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020589041095890408,
+      "loss": 0.1973,
+      "step": 9563
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020587795765877954,
+      "loss": 0.2502,
+      "step": 9564
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020586550435865503,
+      "loss": 0.348,
+      "step": 9565
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0002058530510585305,
+      "loss": 0.3997,
+      "step": 9566
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020584059775840597,
+      "loss": 0.1621,
+      "step": 9567
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0002058281444582814,
+      "loss": 0.3186,
+      "step": 9568
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0002058156911581569,
+      "loss": 0.1546,
+      "step": 9569
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020580323785803235,
+      "loss": 0.1865,
+      "step": 9570
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020579078455790784,
+      "loss": 0.4665,
+      "step": 9571
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0002057783312577833,
+      "loss": 0.3927,
+      "step": 9572
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020576587795765878,
+      "loss": 0.1612,
+      "step": 9573
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020575342465753421,
+      "loss": 0.2293,
+      "step": 9574
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0002057409713574097,
+      "loss": 0.2563,
+      "step": 9575
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020572851805728516,
+      "loss": 0.2892,
+      "step": 9576
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020571606475716062,
+      "loss": 0.5863,
+      "step": 9577
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0002057036114570361,
+      "loss": 0.2815,
+      "step": 9578
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020569115815691154,
+      "loss": 0.2328,
+      "step": 9579
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020567870485678702,
+      "loss": 0.2425,
+      "step": 9580
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020566625155666248,
+      "loss": 0.2993,
+      "step": 9581
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020565379825653797,
+      "loss": 0.2341,
+      "step": 9582
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020564134495641343,
+      "loss": 0.243,
+      "step": 9583
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020562889165628891,
+      "loss": 0.2773,
+      "step": 9584
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020561643835616435,
+      "loss": 0.186,
+      "step": 9585
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020560398505603983,
+      "loss": 0.2684,
+      "step": 9586
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0002055915317559153,
+      "loss": 0.1571,
+      "step": 9587
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00020557907845579078,
+      "loss": 0.2137,
+      "step": 9588
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020556662515566624,
+      "loss": 0.1954,
+      "step": 9589
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020555417185554172,
+      "loss": 0.3659,
+      "step": 9590
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020554171855541716,
+      "loss": 0.2043,
+      "step": 9591
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020552926525529262,
+      "loss": 0.2027,
+      "step": 9592
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0002055168119551681,
+      "loss": 0.1984,
+      "step": 9593
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020550435865504356,
+      "loss": 0.2948,
+      "step": 9594
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020549190535491905,
+      "loss": 0.1865,
+      "step": 9595
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020547945205479448,
+      "loss": 0.3154,
+      "step": 9596
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020546699875466997,
+      "loss": 0.3137,
+      "step": 9597
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020545454545454542,
+      "loss": 0.2956,
+      "step": 9598
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0002054420921544209,
+      "loss": 0.15,
+      "step": 9599
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020542963885429637,
+      "loss": 0.1985,
+      "step": 9600
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020541718555417186,
+      "loss": 0.1588,
+      "step": 9601
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0002054047322540473,
+      "loss": 0.4259,
+      "step": 9602
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020539227895392277,
+      "loss": 0.1853,
+      "step": 9603
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020537982565379823,
+      "loss": 0.1925,
+      "step": 9604
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020536737235367372,
+      "loss": 0.1778,
+      "step": 9605
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020535491905354918,
+      "loss": 0.1947,
+      "step": 9606
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0002053424657534246,
+      "loss": 0.2171,
+      "step": 9607
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0002053300124533001,
+      "loss": 0.2487,
+      "step": 9608
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020531755915317556,
+      "loss": 0.3407,
+      "step": 9609
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020530510585305104,
+      "loss": 0.344,
+      "step": 9610
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0002052926525529265,
+      "loss": 0.1779,
+      "step": 9611
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.000205280199252802,
+      "loss": 0.343,
+      "step": 9612
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00020526774595267742,
+      "loss": 0.1866,
+      "step": 9613
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0002052552926525529,
+      "loss": 0.1973,
+      "step": 9614
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020524283935242837,
+      "loss": 0.41,
+      "step": 9615
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020523038605230385,
+      "loss": 0.2826,
+      "step": 9616
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0002052179327521793,
+      "loss": 0.2623,
+      "step": 9617
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0002052054794520548,
+      "loss": 0.2673,
+      "step": 9618
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020519302615193023,
+      "loss": 0.1754,
+      "step": 9619
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020518057285180572,
+      "loss": 0.2868,
+      "step": 9620
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020516811955168117,
+      "loss": 0.2647,
+      "step": 9621
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020515566625155666,
+      "loss": 0.2122,
+      "step": 9622
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020514321295143212,
+      "loss": 0.1849,
+      "step": 9623
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020513075965130755,
+      "loss": 0.2141,
+      "step": 9624
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020511830635118304,
+      "loss": 0.1606,
+      "step": 9625
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0002051058530510585,
+      "loss": 0.2793,
+      "step": 9626
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020509339975093398,
+      "loss": 0.4279,
+      "step": 9627
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020508094645080944,
+      "loss": 0.227,
+      "step": 9628
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020506849315068493,
+      "loss": 0.216,
+      "step": 9629
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020505603985056036,
+      "loss": 0.2111,
+      "step": 9630
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020504358655043585,
+      "loss": 0.1709,
+      "step": 9631
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0002050311332503113,
+      "loss": 0.224,
+      "step": 9632
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0002050186799501868,
+      "loss": 0.2434,
+      "step": 9633
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020500622665006225,
+      "loss": 0.3616,
+      "step": 9634
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020499377334993774,
+      "loss": 0.1738,
+      "step": 9635
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020498132004981317,
+      "loss": 0.2541,
+      "step": 9636
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020496886674968866,
+      "loss": 0.1939,
+      "step": 9637
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020495641344956412,
+      "loss": 0.2163,
+      "step": 9638
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020494396014943958,
+      "loss": 0.1752,
+      "step": 9639
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00020493150684931506,
+      "loss": 0.3503,
+      "step": 9640
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0002049190535491905,
+      "loss": 0.3971,
+      "step": 9641
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020490660024906598,
+      "loss": 0.6254,
+      "step": 9642
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020489414694894144,
+      "loss": 0.1838,
+      "step": 9643
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020488169364881693,
+      "loss": 0.2567,
+      "step": 9644
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020486924034869238,
+      "loss": 0.1864,
+      "step": 9645
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020485678704856787,
+      "loss": 0.5407,
+      "step": 9646
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0002048443337484433,
+      "loss": 0.2884,
+      "step": 9647
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0002048318804483188,
+      "loss": 0.1759,
+      "step": 9648
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020481942714819425,
+      "loss": 0.1557,
+      "step": 9649
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020480697384806973,
+      "loss": 0.2599,
+      "step": 9650
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0002047945205479452,
+      "loss": 0.2562,
+      "step": 9651
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020478206724782068,
+      "loss": 0.2884,
+      "step": 9652
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0002047696139476961,
+      "loss": 0.2346,
+      "step": 9653
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020475716064757157,
+      "loss": 0.2121,
+      "step": 9654
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020474470734744706,
+      "loss": 0.1953,
+      "step": 9655
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020473225404732252,
+      "loss": 0.2262,
+      "step": 9656
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.000204719800747198,
+      "loss": 0.2106,
+      "step": 9657
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020470734744707344,
+      "loss": 0.3632,
+      "step": 9658
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020469489414694892,
+      "loss": 0.2965,
+      "step": 9659
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020468244084682438,
+      "loss": 0.4185,
+      "step": 9660
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020466998754669987,
+      "loss": 0.2349,
+      "step": 9661
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020465753424657533,
+      "loss": 0.3896,
+      "step": 9662
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0002046450809464508,
+      "loss": 0.2671,
+      "step": 9663
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020463262764632624,
+      "loss": 0.2759,
+      "step": 9664
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00020462017434620173,
+      "loss": 0.5071,
+      "step": 9665
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0002046077210460772,
+      "loss": 0.2245,
+      "step": 9666
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020459526774595268,
+      "loss": 0.2618,
+      "step": 9667
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020458281444582814,
+      "loss": 0.189,
+      "step": 9668
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020457036114570362,
+      "loss": 0.2179,
+      "step": 9669
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020455790784557905,
+      "loss": 0.2363,
+      "step": 9670
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0002045454545454545,
+      "loss": 0.4157,
+      "step": 9671
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020453300124533,
+      "loss": 0.1916,
+      "step": 9672
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020452054794520546,
+      "loss": 0.1776,
+      "step": 9673
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020450809464508094,
+      "loss": 0.1769,
+      "step": 9674
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020449564134495638,
+      "loss": 0.2234,
+      "step": 9675
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020448318804483186,
+      "loss": 0.2524,
+      "step": 9676
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020447073474470732,
+      "loss": 0.3573,
+      "step": 9677
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0002044582814445828,
+      "loss": 0.2175,
+      "step": 9678
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020444582814445827,
+      "loss": 0.328,
+      "step": 9679
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020443337484433375,
+      "loss": 0.1977,
+      "step": 9680
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020442092154420919,
+      "loss": 0.1352,
+      "step": 9681
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020440846824408467,
+      "loss": 0.2239,
+      "step": 9682
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020439601494396013,
+      "loss": 0.2718,
+      "step": 9683
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020438356164383562,
+      "loss": 0.3676,
+      "step": 9684
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020437110834371108,
+      "loss": 0.2722,
+      "step": 9685
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0002043586550435865,
+      "loss": 0.2022,
+      "step": 9686
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.000204346201743462,
+      "loss": 0.2807,
+      "step": 9687
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020433374844333745,
+      "loss": 0.3503,
+      "step": 9688
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020432129514321294,
+      "loss": 0.2219,
+      "step": 9689
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0002043088418430884,
+      "loss": 0.3839,
+      "step": 9690
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020429638854296389,
+      "loss": 0.3582,
+      "step": 9691
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00020428393524283932,
+      "loss": 0.2382,
+      "step": 9692
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0002042714819427148,
+      "loss": 0.2667,
+      "step": 9693
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020425902864259026,
+      "loss": 0.1576,
+      "step": 9694
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020424657534246575,
+      "loss": 0.3279,
+      "step": 9695
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0002042341220423412,
+      "loss": 0.3547,
+      "step": 9696
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0002042216687422167,
+      "loss": 0.3441,
+      "step": 9697
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020420921544209213,
+      "loss": 0.2772,
+      "step": 9698
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0002041967621419676,
+      "loss": 0.2252,
+      "step": 9699
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020418430884184307,
+      "loss": 0.1644,
+      "step": 9700
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020417185554171853,
+      "loss": 0.253,
+      "step": 9701
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020415940224159402,
+      "loss": 0.3803,
+      "step": 9702
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020414694894146945,
+      "loss": 0.3218,
+      "step": 9703
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020413449564134494,
+      "loss": 0.2891,
+      "step": 9704
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0002041220423412204,
+      "loss": 0.1554,
+      "step": 9705
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020410958904109588,
+      "loss": 0.232,
+      "step": 9706
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020409713574097134,
+      "loss": 0.2739,
+      "step": 9707
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020408468244084683,
+      "loss": 0.2812,
+      "step": 9708
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020407222914072226,
+      "loss": 0.3766,
+      "step": 9709
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020405977584059775,
+      "loss": 0.1985,
+      "step": 9710
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0002040473225404732,
+      "loss": 0.2926,
+      "step": 9711
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0002040348692403487,
+      "loss": 0.3455,
+      "step": 9712
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020402241594022415,
+      "loss": 0.1337,
+      "step": 9713
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0002040099626400996,
+      "loss": 0.2407,
+      "step": 9714
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020399750933997507,
+      "loss": 0.4159,
+      "step": 9715
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020398505603985055,
+      "loss": 0.227,
+      "step": 9716
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020397260273972601,
+      "loss": 0.2215,
+      "step": 9717
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00020396014943960147,
+      "loss": 0.2584,
+      "step": 9718
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020394769613947696,
+      "loss": 0.195,
+      "step": 9719
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0002039352428393524,
+      "loss": 0.2833,
+      "step": 9720
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020392278953922788,
+      "loss": 0.3447,
+      "step": 9721
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020391033623910334,
+      "loss": 0.1791,
+      "step": 9722
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020389788293897882,
+      "loss": 0.2497,
+      "step": 9723
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020388542963885428,
+      "loss": 0.201,
+      "step": 9724
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020387297633872974,
+      "loss": 0.2964,
+      "step": 9725
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0002038605230386052,
+      "loss": 0.1765,
+      "step": 9726
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0002038480697384807,
+      "loss": 0.4023,
+      "step": 9727
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020383561643835615,
+      "loss": 0.2328,
+      "step": 9728
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020382316313823163,
+      "loss": 0.2226,
+      "step": 9729
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020381070983810706,
+      "loss": 0.3855,
+      "step": 9730
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020379825653798255,
+      "loss": 0.3071,
+      "step": 9731
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.000203785803237858,
+      "loss": 0.2534,
+      "step": 9732
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020377334993773347,
+      "loss": 0.2992,
+      "step": 9733
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020376089663760896,
+      "loss": 0.3161,
+      "step": 9734
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020374844333748441,
+      "loss": 0.2219,
+      "step": 9735
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020373599003735987,
+      "loss": 0.1843,
+      "step": 9736
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020372353673723533,
+      "loss": 0.1868,
+      "step": 9737
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020371108343711082,
+      "loss": 0.1566,
+      "step": 9738
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020369863013698628,
+      "loss": 0.2091,
+      "step": 9739
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020368617683686176,
+      "loss": 0.4016,
+      "step": 9740
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0002036737235367372,
+      "loss": 0.2385,
+      "step": 9741
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020366127023661268,
+      "loss": 0.2419,
+      "step": 9742
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020364881693648814,
+      "loss": 0.2761,
+      "step": 9743
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00020363636363636363,
+      "loss": 0.1838,
+      "step": 9744
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0002036239103362391,
+      "loss": 0.1966,
+      "step": 9745
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020361145703611457,
+      "loss": 0.3324,
+      "step": 9746
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020359900373599,
+      "loss": 0.1739,
+      "step": 9747
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0002035865504358655,
+      "loss": 0.1825,
+      "step": 9748
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020357409713574095,
+      "loss": 0.203,
+      "step": 9749
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0002035616438356164,
+      "loss": 0.2428,
+      "step": 9750
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0002035491905354919,
+      "loss": 0.1884,
+      "step": 9751
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020353673723536733,
+      "loss": 0.5002,
+      "step": 9752
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020352428393524282,
+      "loss": 0.2942,
+      "step": 9753
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020351183063511827,
+      "loss": 0.3107,
+      "step": 9754
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020349937733499376,
+      "loss": 0.2416,
+      "step": 9755
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020348692403486922,
+      "loss": 0.1595,
+      "step": 9756
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0002034744707347447,
+      "loss": 0.2758,
+      "step": 9757
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020346201743462014,
+      "loss": 0.2564,
+      "step": 9758
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020344956413449562,
+      "loss": 0.447,
+      "step": 9759
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020343711083437108,
+      "loss": 0.2243,
+      "step": 9760
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020342465753424657,
+      "loss": 0.1826,
+      "step": 9761
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020341220423412203,
+      "loss": 0.173,
+      "step": 9762
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020339975093399752,
+      "loss": 0.2745,
+      "step": 9763
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020338729763387295,
+      "loss": 0.1516,
+      "step": 9764
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0002033748443337484,
+      "loss": 0.3243,
+      "step": 9765
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0002033623910336239,
+      "loss": 0.3641,
+      "step": 9766
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020334993773349935,
+      "loss": 0.2335,
+      "step": 9767
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020333748443337484,
+      "loss": 0.2292,
+      "step": 9768
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020332503113325027,
+      "loss": 0.3154,
+      "step": 9769
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00020331257783312576,
+      "loss": 0.2963,
+      "step": 9770
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020330012453300122,
+      "loss": 0.4595,
+      "step": 9771
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002032876712328767,
+      "loss": 0.232,
+      "step": 9772
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020327521793275216,
+      "loss": 0.2242,
+      "step": 9773
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020326276463262765,
+      "loss": 0.22,
+      "step": 9774
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020325031133250308,
+      "loss": 0.2623,
+      "step": 9775
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020323785803237857,
+      "loss": 0.169,
+      "step": 9776
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020322540473225402,
+      "loss": 0.3958,
+      "step": 9777
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002032129514321295,
+      "loss": 0.3021,
+      "step": 9778
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020320049813200497,
+      "loss": 0.2572,
+      "step": 9779
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002031880448318804,
+      "loss": 0.2346,
+      "step": 9780
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002031755915317559,
+      "loss": 0.2087,
+      "step": 9781
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020316313823163135,
+      "loss": 0.1977,
+      "step": 9782
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020315068493150683,
+      "loss": 0.214,
+      "step": 9783
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002031382316313823,
+      "loss": 0.3065,
+      "step": 9784
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020312577833125778,
+      "loss": 0.2667,
+      "step": 9785
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002031133250311332,
+      "loss": 0.2238,
+      "step": 9786
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002031008717310087,
+      "loss": 0.2411,
+      "step": 9787
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020308841843088416,
+      "loss": 0.205,
+      "step": 9788
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020307596513075964,
+      "loss": 0.3248,
+      "step": 9789
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002030635118306351,
+      "loss": 0.3323,
+      "step": 9790
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002030510585305106,
+      "loss": 0.2855,
+      "step": 9791
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020303860523038602,
+      "loss": 0.1684,
+      "step": 9792
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002030261519302615,
+      "loss": 0.5249,
+      "step": 9793
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020301369863013697,
+      "loss": 0.2244,
+      "step": 9794
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00020300124533001245,
+      "loss": 0.2137,
+      "step": 9795
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002029887920298879,
+      "loss": 0.4135,
+      "step": 9796
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020297633872976334,
+      "loss": 0.2126,
+      "step": 9797
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020296388542963883,
+      "loss": 0.2107,
+      "step": 9798
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0002029514321295143,
+      "loss": 0.1625,
+      "step": 9799
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020293897882938978,
+      "loss": 0.2494,
+      "step": 9800
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020292652552926523,
+      "loss": 0.2008,
+      "step": 9801
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020291407222914072,
+      "loss": 0.3586,
+      "step": 9802
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020290161892901615,
+      "loss": 0.2932,
+      "step": 9803
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020288916562889164,
+      "loss": 0.2904,
+      "step": 9804
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0002028767123287671,
+      "loss": 0.1985,
+      "step": 9805
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020286425902864258,
+      "loss": 0.2445,
+      "step": 9806
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020285180572851804,
+      "loss": 0.2026,
+      "step": 9807
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020283935242839353,
+      "loss": 0.3325,
+      "step": 9808
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020282689912826896,
+      "loss": 0.3061,
+      "step": 9809
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020281444582814445,
+      "loss": 0.2582,
+      "step": 9810
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0002028019925280199,
+      "loss": 0.2262,
+      "step": 9811
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020278953922789537,
+      "loss": 0.1661,
+      "step": 9812
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020277708592777085,
+      "loss": 0.2226,
+      "step": 9813
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020276463262764629,
+      "loss": 0.323,
+      "step": 9814
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020275217932752177,
+      "loss": 0.354,
+      "step": 9815
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020273972602739723,
+      "loss": 0.2774,
+      "step": 9816
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020272727272727272,
+      "loss": 0.201,
+      "step": 9817
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020271481942714818,
+      "loss": 0.2562,
+      "step": 9818
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020270236612702366,
+      "loss": 0.2814,
+      "step": 9819
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0002026899128268991,
+      "loss": 0.2063,
+      "step": 9820
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020267745952677458,
+      "loss": 0.3906,
+      "step": 9821
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00020266500622665004,
+      "loss": 0.2153,
+      "step": 9822
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020265255292652553,
+      "loss": 0.2106,
+      "step": 9823
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020264009962640099,
+      "loss": 0.1802,
+      "step": 9824
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020262764632627647,
+      "loss": 0.3837,
+      "step": 9825
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002026151930261519,
+      "loss": 0.1923,
+      "step": 9826
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020260273972602736,
+      "loss": 0.3803,
+      "step": 9827
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020259028642590285,
+      "loss": 0.2419,
+      "step": 9828
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002025778331257783,
+      "loss": 0.4474,
+      "step": 9829
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002025653798256538,
+      "loss": 0.3224,
+      "step": 9830
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020255292652552923,
+      "loss": 0.2033,
+      "step": 9831
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002025404732254047,
+      "loss": 0.3021,
+      "step": 9832
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020252801992528017,
+      "loss": 0.2758,
+      "step": 9833
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020251556662515566,
+      "loss": 0.2839,
+      "step": 9834
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020250311332503112,
+      "loss": 0.182,
+      "step": 9835
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002024906600249066,
+      "loss": 0.2652,
+      "step": 9836
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020247820672478204,
+      "loss": 0.157,
+      "step": 9837
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020246575342465752,
+      "loss": 0.2181,
+      "step": 9838
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020245330012453298,
+      "loss": 0.3343,
+      "step": 9839
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020244084682440847,
+      "loss": 0.483,
+      "step": 9840
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020242839352428393,
+      "loss": 0.2405,
+      "step": 9841
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002024159402241594,
+      "loss": 0.2127,
+      "step": 9842
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020240348692403485,
+      "loss": 0.1548,
+      "step": 9843
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002023910336239103,
+      "loss": 0.2123,
+      "step": 9844
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0002023785803237858,
+      "loss": 0.2319,
+      "step": 9845
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020236612702366125,
+      "loss": 0.5474,
+      "step": 9846
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020235367372353674,
+      "loss": 0.229,
+      "step": 9847
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00020234122042341217,
+      "loss": 0.2274,
+      "step": 9848
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020232876712328765,
+      "loss": 0.2682,
+      "step": 9849
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002023163138231631,
+      "loss": 0.1967,
+      "step": 9850
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002023038605230386,
+      "loss": 0.159,
+      "step": 9851
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020229140722291406,
+      "loss": 0.2859,
+      "step": 9852
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020227895392278955,
+      "loss": 0.2084,
+      "step": 9853
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020226650062266498,
+      "loss": 0.2201,
+      "step": 9854
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020225404732254046,
+      "loss": 0.2206,
+      "step": 9855
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020224159402241592,
+      "loss": 0.2724,
+      "step": 9856
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002022291407222914,
+      "loss": 0.2351,
+      "step": 9857
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020221668742216687,
+      "loss": 0.255,
+      "step": 9858
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002022042341220423,
+      "loss": 0.2763,
+      "step": 9859
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020219178082191779,
+      "loss": 0.2876,
+      "step": 9860
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020217932752179325,
+      "loss": 0.1705,
+      "step": 9861
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020216687422166873,
+      "loss": 0.1861,
+      "step": 9862
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002021544209215442,
+      "loss": 0.3405,
+      "step": 9863
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020214196762141968,
+      "loss": 0.3709,
+      "step": 9864
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002021295143212951,
+      "loss": 0.3765,
+      "step": 9865
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002021170610211706,
+      "loss": 0.2852,
+      "step": 9866
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020210460772104605,
+      "loss": 0.1938,
+      "step": 9867
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020209215442092154,
+      "loss": 0.1764,
+      "step": 9868
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.000202079701120797,
+      "loss": 0.3308,
+      "step": 9869
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002020672478206725,
+      "loss": 0.1964,
+      "step": 9870
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020205479452054792,
+      "loss": 0.5348,
+      "step": 9871
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002020423412204234,
+      "loss": 0.2454,
+      "step": 9872
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020202988792029886,
+      "loss": 0.2601,
+      "step": 9873
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00020201743462017432,
+      "loss": 0.1844,
+      "step": 9874
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0002020049813200498,
+      "loss": 0.1903,
+      "step": 9875
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020199252801992524,
+      "loss": 0.2579,
+      "step": 9876
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020198007471980073,
+      "loss": 0.3506,
+      "step": 9877
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002019676214196762,
+      "loss": 0.3034,
+      "step": 9878
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020195516811955167,
+      "loss": 0.2485,
+      "step": 9879
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020194271481942713,
+      "loss": 0.2468,
+      "step": 9880
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020193026151930262,
+      "loss": 0.1248,
+      "step": 9881
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020191780821917805,
+      "loss": 0.2007,
+      "step": 9882
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020190535491905354,
+      "loss": 0.2357,
+      "step": 9883
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.000201892901618929,
+      "loss": 0.2301,
+      "step": 9884
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020188044831880448,
+      "loss": 0.2931,
+      "step": 9885
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020186799501867994,
+      "loss": 0.3084,
+      "step": 9886
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002018555417185554,
+      "loss": 0.211,
+      "step": 9887
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020184308841843086,
+      "loss": 0.2179,
+      "step": 9888
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020183063511830635,
+      "loss": 0.2366,
+      "step": 9889
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002018181818181818,
+      "loss": 0.2567,
+      "step": 9890
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020180572851805726,
+      "loss": 0.1976,
+      "step": 9891
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020179327521793272,
+      "loss": 0.2185,
+      "step": 9892
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020178082191780818,
+      "loss": 0.1842,
+      "step": 9893
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020176836861768367,
+      "loss": 0.2461,
+      "step": 9894
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020175591531755913,
+      "loss": 0.2717,
+      "step": 9895
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020174346201743461,
+      "loss": 0.5152,
+      "step": 9896
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020173100871731007,
+      "loss": 0.2554,
+      "step": 9897
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020171855541718553,
+      "loss": 0.1804,
+      "step": 9898
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.000201706102117061,
+      "loss": 0.1678,
+      "step": 9899
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020169364881693648,
+      "loss": 0.1735,
+      "step": 9900
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00020168119551681194,
+      "loss": 0.1491,
+      "step": 9901
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020166874221668742,
+      "loss": 0.3267,
+      "step": 9902
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020165628891656286,
+      "loss": 0.2514,
+      "step": 9903
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020164383561643834,
+      "loss": 0.1566,
+      "step": 9904
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002016313823163138,
+      "loss": 0.3008,
+      "step": 9905
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020161892901618926,
+      "loss": 0.2266,
+      "step": 9906
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020160647571606475,
+      "loss": 0.1969,
+      "step": 9907
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002015940224159402,
+      "loss": 0.2604,
+      "step": 9908
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020158156911581567,
+      "loss": 0.4197,
+      "step": 9909
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020156911581569112,
+      "loss": 0.3021,
+      "step": 9910
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002015566625155666,
+      "loss": 0.2309,
+      "step": 9911
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020154420921544207,
+      "loss": 0.2082,
+      "step": 9912
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020153175591531756,
+      "loss": 0.2429,
+      "step": 9913
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.000201519302615193,
+      "loss": 0.2469,
+      "step": 9914
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020150684931506847,
+      "loss": 0.3395,
+      "step": 9915
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020149439601494393,
+      "loss": 0.281,
+      "step": 9916
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020148194271481942,
+      "loss": 0.2715,
+      "step": 9917
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020146948941469488,
+      "loss": 0.2379,
+      "step": 9918
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020145703611457037,
+      "loss": 0.1923,
+      "step": 9919
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002014445828144458,
+      "loss": 0.2615,
+      "step": 9920
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020143212951432126,
+      "loss": 0.4481,
+      "step": 9921
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020141967621419674,
+      "loss": 0.2265,
+      "step": 9922
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002014072229140722,
+      "loss": 0.226,
+      "step": 9923
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002013947696139477,
+      "loss": 0.2652,
+      "step": 9924
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020138231631382312,
+      "loss": 0.2229,
+      "step": 9925
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002013698630136986,
+      "loss": 0.1531,
+      "step": 9926
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00020135740971357407,
+      "loss": 0.3462,
+      "step": 9927
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020134495641344955,
+      "loss": 0.2237,
+      "step": 9928
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.000201332503113325,
+      "loss": 0.1843,
+      "step": 9929
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002013200498132005,
+      "loss": 0.1993,
+      "step": 9930
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020130759651307593,
+      "loss": 0.2436,
+      "step": 9931
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020129514321295142,
+      "loss": 0.185,
+      "step": 9932
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020128268991282687,
+      "loss": 0.3387,
+      "step": 9933
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020127023661270236,
+      "loss": 0.377,
+      "step": 9934
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020125778331257782,
+      "loss": 0.3386,
+      "step": 9935
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002012453300124533,
+      "loss": 0.1953,
+      "step": 9936
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020123287671232874,
+      "loss": 0.3286,
+      "step": 9937
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002012204234122042,
+      "loss": 0.2278,
+      "step": 9938
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020120797011207968,
+      "loss": 0.2266,
+      "step": 9939
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020119551681195514,
+      "loss": 0.3406,
+      "step": 9940
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020118306351183063,
+      "loss": 0.2761,
+      "step": 9941
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020117061021170606,
+      "loss": 0.2558,
+      "step": 9942
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020115815691158155,
+      "loss": 0.2167,
+      "step": 9943
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.000201145703611457,
+      "loss": 0.2861,
+      "step": 9944
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002011332503113325,
+      "loss": 0.1973,
+      "step": 9945
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020112079701120795,
+      "loss": 0.4096,
+      "step": 9946
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020110834371108344,
+      "loss": 0.1932,
+      "step": 9947
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020109589041095887,
+      "loss": 0.2055,
+      "step": 9948
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020108343711083436,
+      "loss": 0.2188,
+      "step": 9949
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020107098381070982,
+      "loss": 0.1912,
+      "step": 9950
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002010585305105853,
+      "loss": 0.2377,
+      "step": 9951
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00020104607721046076,
+      "loss": 0.3733,
+      "step": 9952
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0002010336239103362,
+      "loss": 0.1959,
+      "step": 9953
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020102117061021168,
+      "loss": 0.2653,
+      "step": 9954
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020100871731008714,
+      "loss": 0.2245,
+      "step": 9955
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020099626400996263,
+      "loss": 0.1446,
+      "step": 9956
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020098381070983808,
+      "loss": 0.2008,
+      "step": 9957
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020097135740971357,
+      "loss": 0.2756,
+      "step": 9958
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.000200958904109589,
+      "loss": 0.5931,
+      "step": 9959
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002009464508094645,
+      "loss": 0.1999,
+      "step": 9960
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020093399750933995,
+      "loss": 0.2003,
+      "step": 9961
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020092154420921543,
+      "loss": 0.2027,
+      "step": 9962
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002009090909090909,
+      "loss": 0.1639,
+      "step": 9963
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020089663760896638,
+      "loss": 0.2514,
+      "step": 9964
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002008841843088418,
+      "loss": 0.2576,
+      "step": 9965
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002008717310087173,
+      "loss": 0.2627,
+      "step": 9966
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020085927770859276,
+      "loss": 0.2098,
+      "step": 9967
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020084682440846822,
+      "loss": 0.1703,
+      "step": 9968
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002008343711083437,
+      "loss": 0.1913,
+      "step": 9969
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020082191780821914,
+      "loss": 0.2509,
+      "step": 9970
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020080946450809462,
+      "loss": 0.5792,
+      "step": 9971
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020079701120797008,
+      "loss": 0.2382,
+      "step": 9972
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020078455790784557,
+      "loss": 0.1672,
+      "step": 9973
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020077210460772103,
+      "loss": 0.2384,
+      "step": 9974
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002007596513075965,
+      "loss": 0.1506,
+      "step": 9975
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020074719800747194,
+      "loss": 0.2821,
+      "step": 9976
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020073474470734743,
+      "loss": 0.2985,
+      "step": 9977
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002007222914072229,
+      "loss": 0.2067,
+      "step": 9978
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00020070983810709838,
+      "loss": 0.2016,
+      "step": 9979
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020069738480697384,
+      "loss": 0.2039,
+      "step": 9980
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020068493150684932,
+      "loss": 0.1844,
+      "step": 9981
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020067247820672475,
+      "loss": 0.2185,
+      "step": 9982
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020066002490660024,
+      "loss": 0.2345,
+      "step": 9983
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002006475716064757,
+      "loss": 0.3287,
+      "step": 9984
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020063511830635116,
+      "loss": 0.253,
+      "step": 9985
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020062266500622664,
+      "loss": 0.2062,
+      "step": 9986
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020061021170610208,
+      "loss": 0.1701,
+      "step": 9987
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020059775840597756,
+      "loss": 0.1855,
+      "step": 9988
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020058530510585302,
+      "loss": 0.4128,
+      "step": 9989
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002005728518057285,
+      "loss": 0.3522,
+      "step": 9990
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020056039850560397,
+      "loss": 0.2056,
+      "step": 9991
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020054794520547945,
+      "loss": 0.2848,
+      "step": 9992
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020053549190535489,
+      "loss": 0.2093,
+      "step": 9993
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020052303860523037,
+      "loss": 0.2787,
+      "step": 9994
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020051058530510583,
+      "loss": 0.2476,
+      "step": 9995
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020049813200498132,
+      "loss": 0.3763,
+      "step": 9996
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020048567870485678,
+      "loss": 0.3934,
+      "step": 9997
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020047322540473226,
+      "loss": 0.2453,
+      "step": 9998
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002004607721046077,
+      "loss": 0.1856,
+      "step": 9999
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020044831880448315,
+      "loss": 0.1673,
+      "step": 10000
+    },
+    {
+      "epoch": 3.83,
+      "eval_cer": 0.10836373568488207,
+      "eval_loss": 0.36826929450035095,
+      "eval_runtime": 46.9348,
+      "eval_samples_per_second": 37.35,
+      "eval_steps_per_second": 4.687,
+      "eval_wer": 0.1855908041815631,
+      "step": 10000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020043586550435864,
+      "loss": 0.3227,
+      "step": 10001
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002004234122042341,
+      "loss": 0.3616,
+      "step": 10002
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020041095890410959,
+      "loss": 0.2823,
+      "step": 10003
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00020039850560398502,
+      "loss": 0.3208,
+      "step": 10004
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002003860523038605,
+      "loss": 0.1943,
+      "step": 10005
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020037359900373596,
+      "loss": 0.2201,
+      "step": 10006
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020036114570361145,
+      "loss": 0.1549,
+      "step": 10007
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002003486924034869,
+      "loss": 0.2347,
+      "step": 10008
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002003362391033624,
+      "loss": 0.4417,
+      "step": 10009
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020032378580323783,
+      "loss": 0.2588,
+      "step": 10010
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002003113325031133,
+      "loss": 0.2791,
+      "step": 10011
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020029887920298877,
+      "loss": 0.2523,
+      "step": 10012
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020028642590286426,
+      "loss": 0.2386,
+      "step": 10013
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020027397260273972,
+      "loss": 0.2512,
+      "step": 10014
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002002615193026152,
+      "loss": 0.348,
+      "step": 10015
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020024906600249064,
+      "loss": 0.2326,
+      "step": 10016
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002002366127023661,
+      "loss": 0.2009,
+      "step": 10017
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020022415940224158,
+      "loss": 0.1823,
+      "step": 10018
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020021170610211704,
+      "loss": 0.2062,
+      "step": 10019
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020019925280199253,
+      "loss": 0.1973,
+      "step": 10020
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020018679950186796,
+      "loss": 0.33,
+      "step": 10021
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020017434620174345,
+      "loss": 0.2924,
+      "step": 10022
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002001618929016189,
+      "loss": 0.2299,
+      "step": 10023
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002001494396014944,
+      "loss": 0.3906,
+      "step": 10024
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020013698630136985,
+      "loss": 0.2583,
+      "step": 10025
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020012453300124534,
+      "loss": 0.2329,
+      "step": 10026
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020011207970112077,
+      "loss": 0.4244,
+      "step": 10027
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020009962640099625,
+      "loss": 0.3382,
+      "step": 10028
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020008717310087171,
+      "loss": 0.2644,
+      "step": 10029
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0002000747198007472,
+      "loss": 0.2481,
+      "step": 10030
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00020006226650062266,
+      "loss": 0.2847,
+      "step": 10031
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0002000498132004981,
+      "loss": 0.1826,
+      "step": 10032
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00020003735990037358,
+      "loss": 0.3806,
+      "step": 10033
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00020002490660024904,
+      "loss": 0.4653,
+      "step": 10034
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00020001245330012452,
+      "loss": 0.2116,
+      "step": 10035
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 0.1703,
+      "step": 10036
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019998754669987547,
+      "loss": 0.1817,
+      "step": 10037
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001999750933997509,
+      "loss": 0.1672,
+      "step": 10038
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001999626400996264,
+      "loss": 0.219,
+      "step": 10039
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019995018679950185,
+      "loss": 0.2724,
+      "step": 10040
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019993773349937733,
+      "loss": 0.2086,
+      "step": 10041
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001999252801992528,
+      "loss": 0.2067,
+      "step": 10042
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019991282689912828,
+      "loss": 0.2517,
+      "step": 10043
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001999003735990037,
+      "loss": 0.1522,
+      "step": 10044
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001998879202988792,
+      "loss": 0.2631,
+      "step": 10045
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019987546699875466,
+      "loss": 0.3626,
+      "step": 10046
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019986301369863011,
+      "loss": 0.2503,
+      "step": 10047
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001998505603985056,
+      "loss": 0.2548,
+      "step": 10048
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019983810709838103,
+      "loss": 0.2091,
+      "step": 10049
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019982565379825652,
+      "loss": 0.1739,
+      "step": 10050
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019981320049813198,
+      "loss": 0.3632,
+      "step": 10051
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019980074719800746,
+      "loss": 0.3072,
+      "step": 10052
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019978829389788292,
+      "loss": 0.328,
+      "step": 10053
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019977584059775838,
+      "loss": 0.2301,
+      "step": 10054
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019976338729763384,
+      "loss": 0.2911,
+      "step": 10055
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00019975093399750933,
+      "loss": 0.323,
+      "step": 10056
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001997384806973848,
+      "loss": 0.1945,
+      "step": 10057
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019972602739726027,
+      "loss": 0.4137,
+      "step": 10058
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019971357409713573,
+      "loss": 0.4365,
+      "step": 10059
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001997011207970112,
+      "loss": 0.1828,
+      "step": 10060
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019968866749688665,
+      "loss": 0.2251,
+      "step": 10061
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019967621419676214,
+      "loss": 0.1799,
+      "step": 10062
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001996637608966376,
+      "loss": 0.2538,
+      "step": 10063
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019965130759651306,
+      "loss": 0.1913,
+      "step": 10064
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019963885429638852,
+      "loss": 0.338,
+      "step": 10065
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019962640099626397,
+      "loss": 0.4188,
+      "step": 10066
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019961394769613946,
+      "loss": 0.24,
+      "step": 10067
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019960149439601492,
+      "loss": 0.3027,
+      "step": 10068
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001995890410958904,
+      "loss": 0.1748,
+      "step": 10069
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019957658779576587,
+      "loss": 0.2673,
+      "step": 10070
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019956413449564132,
+      "loss": 0.3181,
+      "step": 10071
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019955168119551678,
+      "loss": 0.2607,
+      "step": 10072
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019953922789539227,
+      "loss": 0.2268,
+      "step": 10073
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019952677459526773,
+      "loss": 0.1927,
+      "step": 10074
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019951432129514322,
+      "loss": 0.1885,
+      "step": 10075
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019950186799501865,
+      "loss": 0.2696,
+      "step": 10076
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019948941469489413,
+      "loss": 0.3814,
+      "step": 10077
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001994769613947696,
+      "loss": 0.2569,
+      "step": 10078
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019946450809464505,
+      "loss": 0.2345,
+      "step": 10079
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019945205479452054,
+      "loss": 0.1703,
+      "step": 10080
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.000199439601494396,
+      "loss": 0.2353,
+      "step": 10081
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019942714819427146,
+      "loss": 0.2965,
+      "step": 10082
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00019941469489414692,
+      "loss": 0.2585,
+      "step": 10083
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001994022415940224,
+      "loss": 0.3974,
+      "step": 10084
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019938978829389786,
+      "loss": 0.2238,
+      "step": 10085
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019937733499377335,
+      "loss": 0.1883,
+      "step": 10086
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019936488169364878,
+      "loss": 0.2057,
+      "step": 10087
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019935242839352427,
+      "loss": 0.2167,
+      "step": 10088
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019933997509339972,
+      "loss": 0.2254,
+      "step": 10089
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001993275217932752,
+      "loss": 0.2709,
+      "step": 10090
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019931506849315067,
+      "loss": 0.2246,
+      "step": 10091
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019930261519302616,
+      "loss": 0.2058,
+      "step": 10092
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001992901618929016,
+      "loss": 0.1275,
+      "step": 10093
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019927770859277705,
+      "loss": 0.2962,
+      "step": 10094
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019926525529265253,
+      "loss": 0.2006,
+      "step": 10095
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.000199252801992528,
+      "loss": 0.3745,
+      "step": 10096
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019924034869240348,
+      "loss": 0.1801,
+      "step": 10097
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001992278953922789,
+      "loss": 0.2527,
+      "step": 10098
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001992154420921544,
+      "loss": 0.2524,
+      "step": 10099
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019920298879202986,
+      "loss": 0.3626,
+      "step": 10100
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019919053549190534,
+      "loss": 0.1662,
+      "step": 10101
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001991780821917808,
+      "loss": 0.5867,
+      "step": 10102
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001991656288916563,
+      "loss": 2.5305,
+      "step": 10103
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019915317559153172,
+      "loss": 0.2111,
+      "step": 10104
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001991407222914072,
+      "loss": 0.2568,
+      "step": 10105
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019912826899128267,
+      "loss": 0.3464,
+      "step": 10106
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00019911581569115815,
+      "loss": 0.1852,
+      "step": 10107
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001991033623910336,
+      "loss": 0.3542,
+      "step": 10108
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001990909090909091,
+      "loss": 0.3311,
+      "step": 10109
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019907845579078453,
+      "loss": 0.2311,
+      "step": 10110
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019906600249066,
+      "loss": 0.2158,
+      "step": 10111
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019905354919053548,
+      "loss": 0.1997,
+      "step": 10112
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019904109589041093,
+      "loss": 0.2383,
+      "step": 10113
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019902864259028642,
+      "loss": 0.2689,
+      "step": 10114
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019901618929016185,
+      "loss": 0.517,
+      "step": 10115
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019900373599003734,
+      "loss": 0.3035,
+      "step": 10116
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0001989912826899128,
+      "loss": 0.1499,
+      "step": 10117
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019897882938978828,
+      "loss": 0.3125,
+      "step": 10118
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019896637608966374,
+      "loss": 0.1673,
+      "step": 10119
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019895392278953923,
+      "loss": 0.3444,
+      "step": 10120
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019894146948941466,
+      "loss": 0.3471,
+      "step": 10121
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019892901618929015,
+      "loss": 0.3758,
+      "step": 10122
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0001989165628891656,
+      "loss": 0.2128,
+      "step": 10123
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0001989041095890411,
+      "loss": 0.1702,
+      "step": 10124
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019889165628891655,
+      "loss": 0.2219,
+      "step": 10125
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019887920298879199,
+      "loss": 0.2786,
+      "step": 10126
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019886674968866747,
+      "loss": 0.3075,
+      "step": 10127
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019885429638854293,
+      "loss": 0.2952,
+      "step": 10128
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019884184308841842,
+      "loss": 0.1797,
+      "step": 10129
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019882938978829388,
+      "loss": 0.2363,
+      "step": 10130
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019881693648816936,
+      "loss": 0.2271,
+      "step": 10131
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0001988044831880448,
+      "loss": 0.2915,
+      "step": 10132
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019879202988792028,
+      "loss": 0.2744,
+      "step": 10133
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019877957658779574,
+      "loss": 0.2942,
+      "step": 10134
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00019876712328767123,
+      "loss": 0.2356,
+      "step": 10135
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019875466998754669,
+      "loss": 0.184,
+      "step": 10136
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019874221668742217,
+      "loss": 0.2033,
+      "step": 10137
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001987297633872976,
+      "loss": 0.2361,
+      "step": 10138
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001987173100871731,
+      "loss": 0.2239,
+      "step": 10139
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019870485678704855,
+      "loss": 0.3394,
+      "step": 10140
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.000198692403486924,
+      "loss": 0.3032,
+      "step": 10141
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001986799501867995,
+      "loss": 0.2216,
+      "step": 10142
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019866749688667493,
+      "loss": 0.3197,
+      "step": 10143
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001986550435865504,
+      "loss": 0.2212,
+      "step": 10144
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019864259028642587,
+      "loss": 0.2172,
+      "step": 10145
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019863013698630136,
+      "loss": 0.5422,
+      "step": 10146
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019861768368617682,
+      "loss": 0.2491,
+      "step": 10147
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001986052303860523,
+      "loss": 0.2005,
+      "step": 10148
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019859277708592774,
+      "loss": 0.1866,
+      "step": 10149
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019858032378580322,
+      "loss": 0.2705,
+      "step": 10150
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019856787048567868,
+      "loss": 0.2015,
+      "step": 10151
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019855541718555417,
+      "loss": 0.2852,
+      "step": 10152
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019854296388542963,
+      "loss": 0.2154,
+      "step": 10153
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001985305105853051,
+      "loss": 0.2158,
+      "step": 10154
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019851805728518055,
+      "loss": 0.4618,
+      "step": 10155
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019850560398505603,
+      "loss": 0.2556,
+      "step": 10156
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001984931506849315,
+      "loss": 0.2138,
+      "step": 10157
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019848069738480695,
+      "loss": 0.2031,
+      "step": 10158
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019846824408468244,
+      "loss": 0.4201,
+      "step": 10159
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019845579078455787,
+      "loss": 0.3185,
+      "step": 10160
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00019844333748443335,
+      "loss": 0.2335,
+      "step": 10161
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001984308841843088,
+      "loss": 0.1579,
+      "step": 10162
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0001984184308841843,
+      "loss": 0.3726,
+      "step": 10163
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019840597758405976,
+      "loss": 0.2294,
+      "step": 10164
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019839352428393525,
+      "loss": 0.3527,
+      "step": 10165
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019838107098381068,
+      "loss": 0.193,
+      "step": 10166
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019836861768368616,
+      "loss": 0.2244,
+      "step": 10167
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019835616438356162,
+      "loss": 0.1497,
+      "step": 10168
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0001983437110834371,
+      "loss": 0.2085,
+      "step": 10169
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019833125778331257,
+      "loss": 0.2909,
+      "step": 10170
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019831880448318805,
+      "loss": 0.3075,
+      "step": 10171
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0001983063511830635,
+      "loss": 0.1993,
+      "step": 10172
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019829389788293895,
+      "loss": 0.4578,
+      "step": 10173
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019828144458281443,
+      "loss": 0.2671,
+      "step": 10174
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0001982689912826899,
+      "loss": 0.2392,
+      "step": 10175
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019825653798256538,
+      "loss": 0.1474,
+      "step": 10176
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0001982440846824408,
+      "loss": 0.3722,
+      "step": 10177
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0001982316313823163,
+      "loss": 0.2947,
+      "step": 10178
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019821917808219175,
+      "loss": 0.4376,
+      "step": 10179
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019820672478206724,
+      "loss": 0.1796,
+      "step": 10180
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0001981942714819427,
+      "loss": 0.3243,
+      "step": 10181
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0001981818181818182,
+      "loss": 0.2258,
+      "step": 10182
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019816936488169362,
+      "loss": 0.3358,
+      "step": 10183
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0001981569115815691,
+      "loss": 0.416,
+      "step": 10184
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019814445828144456,
+      "loss": 0.2335,
+      "step": 10185
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019813200498132005,
+      "loss": 0.269,
+      "step": 10186
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0001981195516811955,
+      "loss": 0.3855,
+      "step": 10187
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00019810709838107094,
+      "loss": 0.2104,
+      "step": 10188
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019809464508094643,
+      "loss": 0.2114,
+      "step": 10189
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0001980821917808219,
+      "loss": 0.3783,
+      "step": 10190
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019806973848069737,
+      "loss": 0.2041,
+      "step": 10191
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019805728518057283,
+      "loss": 0.5896,
+      "step": 10192
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019804483188044832,
+      "loss": 0.2129,
+      "step": 10193
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019803237858032375,
+      "loss": 0.1704,
+      "step": 10194
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019801992528019924,
+      "loss": 0.2406,
+      "step": 10195
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0001980074719800747,
+      "loss": 0.2411,
+      "step": 10196
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019799501867995018,
+      "loss": 0.2849,
+      "step": 10197
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019798256537982564,
+      "loss": 0.2243,
+      "step": 10198
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019797011207970113,
+      "loss": 0.2027,
+      "step": 10199
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019795765877957656,
+      "loss": 0.2664,
+      "step": 10200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019794520547945205,
+      "loss": 0.1407,
+      "step": 10201
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0001979327521793275,
+      "loss": 0.262,
+      "step": 10202
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.000197920298879203,
+      "loss": 0.2923,
+      "step": 10203
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019790784557907845,
+      "loss": 0.1854,
+      "step": 10204
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019789539227895388,
+      "loss": 0.1728,
+      "step": 10205
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019788293897882937,
+      "loss": 0.1965,
+      "step": 10206
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019787048567870483,
+      "loss": 0.2065,
+      "step": 10207
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019785803237858031,
+      "loss": 0.2074,
+      "step": 10208
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019784557907845577,
+      "loss": 0.4217,
+      "step": 10209
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019783312577833126,
+      "loss": 0.2113,
+      "step": 10210
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0001978206724782067,
+      "loss": 0.215,
+      "step": 10211
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019780821917808218,
+      "loss": 0.3062,
+      "step": 10212
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019779576587795764,
+      "loss": 0.2335,
+      "step": 10213
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00019778331257783312,
+      "loss": 0.1998,
+      "step": 10214
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019777085927770858,
+      "loss": 0.4551,
+      "step": 10215
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019775840597758404,
+      "loss": 0.2304,
+      "step": 10216
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0001977459526774595,
+      "loss": 0.2599,
+      "step": 10217
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.000197733499377335,
+      "loss": 0.1871,
+      "step": 10218
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019772104607721045,
+      "loss": 0.2645,
+      "step": 10219
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0001977085927770859,
+      "loss": 0.3118,
+      "step": 10220
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0001976961394769614,
+      "loss": 0.4373,
+      "step": 10221
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019768368617683682,
+      "loss": 0.2511,
+      "step": 10222
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0001976712328767123,
+      "loss": 0.2401,
+      "step": 10223
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019765877957658777,
+      "loss": 0.2008,
+      "step": 10224
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019764632627646326,
+      "loss": 0.1811,
+      "step": 10225
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019763387297633872,
+      "loss": 0.1669,
+      "step": 10226
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019762141967621417,
+      "loss": 0.437,
+      "step": 10227
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019760896637608963,
+      "loss": 0.2162,
+      "step": 10228
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019759651307596512,
+      "loss": 0.2709,
+      "step": 10229
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019758405977584058,
+      "loss": 0.2873,
+      "step": 10230
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019757160647571607,
+      "loss": 0.2858,
+      "step": 10231
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019755915317559152,
+      "loss": 0.1927,
+      "step": 10232
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019754669987546698,
+      "loss": 0.2232,
+      "step": 10233
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019753424657534244,
+      "loss": 0.314,
+      "step": 10234
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0001975217932752179,
+      "loss": 0.2489,
+      "step": 10235
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0001975093399750934,
+      "loss": 0.202,
+      "step": 10236
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019749688667496885,
+      "loss": 0.1552,
+      "step": 10237
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.0001974844333748443,
+      "loss": 0.3602,
+      "step": 10238
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019747198007471977,
+      "loss": 0.3223,
+      "step": 10239
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00019745952677459525,
+      "loss": 0.3714,
+      "step": 10240
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0001974470734744707,
+      "loss": 0.2227,
+      "step": 10241
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0001974346201743462,
+      "loss": 0.2024,
+      "step": 10242
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019742216687422166,
+      "loss": 0.1955,
+      "step": 10243
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019740971357409712,
+      "loss": 0.1705,
+      "step": 10244
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019739726027397258,
+      "loss": 0.2396,
+      "step": 10245
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019738480697384806,
+      "loss": 0.2885,
+      "step": 10246
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019737235367372352,
+      "loss": 0.2365,
+      "step": 10247
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.000197359900373599,
+      "loss": 0.457,
+      "step": 10248
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019734744707347444,
+      "loss": 0.2616,
+      "step": 10249
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019733499377334993,
+      "loss": 0.1649,
+      "step": 10250
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019732254047322538,
+      "loss": 0.1977,
+      "step": 10251
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019731008717310084,
+      "loss": 0.3508,
+      "step": 10252
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019729763387297633,
+      "loss": 0.244,
+      "step": 10253
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0001972851805728518,
+      "loss": 0.3226,
+      "step": 10254
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019727272727272725,
+      "loss": 0.1862,
+      "step": 10255
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0001972602739726027,
+      "loss": 0.1247,
+      "step": 10256
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0001972478206724782,
+      "loss": 0.1997,
+      "step": 10257
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019723536737235365,
+      "loss": 0.1891,
+      "step": 10258
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019722291407222914,
+      "loss": 0.4432,
+      "step": 10259
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019721046077210457,
+      "loss": 0.2012,
+      "step": 10260
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019719800747198006,
+      "loss": 0.2534,
+      "step": 10261
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019718555417185552,
+      "loss": 0.2823,
+      "step": 10262
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.000197173100871731,
+      "loss": 0.2295,
+      "step": 10263
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019716064757160646,
+      "loss": 0.2084,
+      "step": 10264
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019714819427148195,
+      "loss": 0.2807,
+      "step": 10265
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00019713574097135738,
+      "loss": 0.1855,
+      "step": 10266
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019712328767123284,
+      "loss": 0.2061,
+      "step": 10267
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019711083437110833,
+      "loss": 0.2506,
+      "step": 10268
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019709838107098378,
+      "loss": 0.1719,
+      "step": 10269
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019708592777085927,
+      "loss": 0.1624,
+      "step": 10270
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0001970734744707347,
+      "loss": 0.5005,
+      "step": 10271
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0001970610211706102,
+      "loss": 0.1998,
+      "step": 10272
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019704856787048565,
+      "loss": 0.1767,
+      "step": 10273
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019703611457036113,
+      "loss": 0.1832,
+      "step": 10274
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0001970236612702366,
+      "loss": 0.1622,
+      "step": 10275
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019701120797011208,
+      "loss": 0.2045,
+      "step": 10276
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0001969987546699875,
+      "loss": 0.4421,
+      "step": 10277
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.000196986301369863,
+      "loss": 0.2046,
+      "step": 10278
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019697384806973846,
+      "loss": 0.1995,
+      "step": 10279
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019696139476961394,
+      "loss": 0.2632,
+      "step": 10280
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0001969489414694894,
+      "loss": 0.2128,
+      "step": 10281
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0001969364881693649,
+      "loss": 0.2242,
+      "step": 10282
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019692403486924032,
+      "loss": 0.3466,
+      "step": 10283
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019691158156911578,
+      "loss": 0.4143,
+      "step": 10284
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019689912826899127,
+      "loss": 0.2223,
+      "step": 10285
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019688667496886673,
+      "loss": 0.232,
+      "step": 10286
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0001968742216687422,
+      "loss": 0.1791,
+      "step": 10287
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019686176836861764,
+      "loss": 0.259,
+      "step": 10288
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019684931506849313,
+      "loss": 0.2335,
+      "step": 10289
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0001968368617683686,
+      "loss": 0.3218,
+      "step": 10290
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019682440846824408,
+      "loss": 0.2576,
+      "step": 10291
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00019681195516811954,
+      "loss": 0.2717,
+      "step": 10292
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019679950186799502,
+      "loss": 0.153,
+      "step": 10293
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019678704856787045,
+      "loss": 0.1995,
+      "step": 10294
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019677459526774594,
+      "loss": 0.3557,
+      "step": 10295
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0001967621419676214,
+      "loss": 0.4074,
+      "step": 10296
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019674968866749689,
+      "loss": 0.3461,
+      "step": 10297
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019673723536737234,
+      "loss": 0.1823,
+      "step": 10298
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019672478206724778,
+      "loss": 0.1845,
+      "step": 10299
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019671232876712326,
+      "loss": 0.2,
+      "step": 10300
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019669987546699872,
+      "loss": 0.2348,
+      "step": 10301
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0001966874221668742,
+      "loss": 0.4469,
+      "step": 10302
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019667496886674967,
+      "loss": 0.2145,
+      "step": 10303
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019666251556662515,
+      "loss": 0.2536,
+      "step": 10304
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019665006226650059,
+      "loss": 0.2909,
+      "step": 10305
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019663760896637607,
+      "loss": 0.2412,
+      "step": 10306
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019662515566625153,
+      "loss": 0.2402,
+      "step": 10307
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019661270236612702,
+      "loss": 0.2884,
+      "step": 10308
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019660024906600248,
+      "loss": 0.27,
+      "step": 10309
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019658779576587796,
+      "loss": 0.3211,
+      "step": 10310
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0001965753424657534,
+      "loss": 0.3043,
+      "step": 10311
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019656288916562888,
+      "loss": 0.3431,
+      "step": 10312
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019655043586550434,
+      "loss": 0.3342,
+      "step": 10313
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0001965379825653798,
+      "loss": 0.2212,
+      "step": 10314
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019652552926525529,
+      "loss": 0.4373,
+      "step": 10315
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019651307596513072,
+      "loss": 0.3269,
+      "step": 10316
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.0001965006226650062,
+      "loss": 0.2581,
+      "step": 10317
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00019648816936488166,
+      "loss": 0.2535,
+      "step": 10318
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019647571606475715,
+      "loss": 0.1459,
+      "step": 10319
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0001964632627646326,
+      "loss": 0.1818,
+      "step": 10320
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0001964508094645081,
+      "loss": 0.5125,
+      "step": 10321
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019643835616438353,
+      "loss": 0.2939,
+      "step": 10322
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019642590286425901,
+      "loss": 0.2959,
+      "step": 10323
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019641344956413447,
+      "loss": 0.2794,
+      "step": 10324
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019640099626400996,
+      "loss": 0.195,
+      "step": 10325
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019638854296388542,
+      "loss": 0.21,
+      "step": 10326
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0001963760896637609,
+      "loss": 0.3439,
+      "step": 10327
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019636363636363634,
+      "loss": 0.2621,
+      "step": 10328
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019635118306351182,
+      "loss": 0.2962,
+      "step": 10329
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019633872976338728,
+      "loss": 0.215,
+      "step": 10330
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019632627646326274,
+      "loss": 0.1804,
+      "step": 10331
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019631382316313823,
+      "loss": 0.2311,
+      "step": 10332
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019630136986301366,
+      "loss": 0.18,
+      "step": 10333
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019628891656288915,
+      "loss": 0.257,
+      "step": 10334
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0001962764632627646,
+      "loss": 0.2366,
+      "step": 10335
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0001962640099626401,
+      "loss": 0.1964,
+      "step": 10336
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019625155666251555,
+      "loss": 0.144,
+      "step": 10337
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019623910336239104,
+      "loss": 0.2068,
+      "step": 10338
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019622665006226647,
+      "loss": 0.2729,
+      "step": 10339
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019621419676214196,
+      "loss": 0.3251,
+      "step": 10340
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019620174346201741,
+      "loss": 0.386,
+      "step": 10341
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0001961892901618929,
+      "loss": 0.275,
+      "step": 10342
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019617683686176836,
+      "loss": 0.2216,
+      "step": 10343
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00019616438356164385,
+      "loss": 0.2201,
+      "step": 10344
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019615193026151928,
+      "loss": 0.2148,
+      "step": 10345
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019613947696139474,
+      "loss": 0.3678,
+      "step": 10346
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019612702366127022,
+      "loss": 0.2857,
+      "step": 10347
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019611457036114568,
+      "loss": 0.246,
+      "step": 10348
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019610211706102117,
+      "loss": 0.1876,
+      "step": 10349
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0001960896637608966,
+      "loss": 0.2839,
+      "step": 10350
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0001960772104607721,
+      "loss": 0.2138,
+      "step": 10351
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019606475716064755,
+      "loss": 0.3427,
+      "step": 10352
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019605230386052303,
+      "loss": 0.2406,
+      "step": 10353
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0001960398505603985,
+      "loss": 0.2727,
+      "step": 10354
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019602739726027398,
+      "loss": 0.2617,
+      "step": 10355
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0001960149439601494,
+      "loss": 0.2199,
+      "step": 10356
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0001960024906600249,
+      "loss": 0.1621,
+      "step": 10357
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019599003735990036,
+      "loss": 0.239,
+      "step": 10358
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019597758405977584,
+      "loss": 0.352,
+      "step": 10359
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0001959651307596513,
+      "loss": 0.463,
+      "step": 10360
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019595267745952673,
+      "loss": 0.2315,
+      "step": 10361
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019594022415940222,
+      "loss": 0.326,
+      "step": 10362
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019592777085927768,
+      "loss": 0.2172,
+      "step": 10363
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019591531755915316,
+      "loss": 0.3342,
+      "step": 10364
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019590286425902862,
+      "loss": 0.3407,
+      "step": 10365
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0001958904109589041,
+      "loss": 0.285,
+      "step": 10366
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019587795765877954,
+      "loss": 0.2591,
+      "step": 10367
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019586550435865503,
+      "loss": 0.1903,
+      "step": 10368
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0001958530510585305,
+      "loss": 0.1697,
+      "step": 10369
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00019584059775840597,
+      "loss": 0.1849,
+      "step": 10370
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019582814445828143,
+      "loss": 0.4605,
+      "step": 10371
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019581569115815692,
+      "loss": 0.2393,
+      "step": 10372
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019580323785803235,
+      "loss": 0.2978,
+      "step": 10373
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019579078455790784,
+      "loss": 0.2071,
+      "step": 10374
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0001957783312577833,
+      "loss": 0.1752,
+      "step": 10375
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019576587795765878,
+      "loss": 0.2094,
+      "step": 10376
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019575342465753424,
+      "loss": 0.4598,
+      "step": 10377
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019574097135740967,
+      "loss": 0.2051,
+      "step": 10378
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019572851805728516,
+      "loss": 0.2459,
+      "step": 10379
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019571606475716062,
+      "loss": 0.2039,
+      "step": 10380
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0001957036114570361,
+      "loss": 0.1715,
+      "step": 10381
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019569115815691157,
+      "loss": 0.1886,
+      "step": 10382
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019567870485678705,
+      "loss": 0.1915,
+      "step": 10383
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019566625155666248,
+      "loss": 0.2679,
+      "step": 10384
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019565379825653797,
+      "loss": 0.3995,
+      "step": 10385
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019564134495641343,
+      "loss": 0.2373,
+      "step": 10386
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019562889165628892,
+      "loss": 0.1729,
+      "step": 10387
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019561643835616437,
+      "loss": 0.1703,
+      "step": 10388
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019560398505603983,
+      "loss": 0.24,
+      "step": 10389
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0001955915317559153,
+      "loss": 0.51,
+      "step": 10390
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019557907845579078,
+      "loss": 0.1801,
+      "step": 10391
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019556662515566624,
+      "loss": 0.1878,
+      "step": 10392
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0001955541718555417,
+      "loss": 0.2179,
+      "step": 10393
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019554171855541718,
+      "loss": 0.1796,
+      "step": 10394
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00019552926525529262,
+      "loss": 0.1743,
+      "step": 10395
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0001955168119551681,
+      "loss": 0.3555,
+      "step": 10396
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019550435865504356,
+      "loss": 0.2243,
+      "step": 10397
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019549190535491905,
+      "loss": 0.1862,
+      "step": 10398
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001954794520547945,
+      "loss": 0.195,
+      "step": 10399
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019546699875466997,
+      "loss": 0.2521,
+      "step": 10400
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019545454545454543,
+      "loss": 0.1873,
+      "step": 10401
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001954420921544209,
+      "loss": 0.3768,
+      "step": 10402
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019542963885429637,
+      "loss": 0.2241,
+      "step": 10403
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019541718555417186,
+      "loss": 0.2138,
+      "step": 10404
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019540473225404732,
+      "loss": 0.2042,
+      "step": 10405
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019539227895392278,
+      "loss": 0.215,
+      "step": 10406
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019537982565379823,
+      "loss": 0.3001,
+      "step": 10407
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001953673723536737,
+      "loss": 0.1994,
+      "step": 10408
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019535491905354918,
+      "loss": 0.1949,
+      "step": 10409
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019534246575342464,
+      "loss": 0.3013,
+      "step": 10410
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001953300124533001,
+      "loss": 0.2341,
+      "step": 10411
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019531755915317556,
+      "loss": 0.171,
+      "step": 10412
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019530510585305104,
+      "loss": 0.251,
+      "step": 10413
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001952926525529265,
+      "loss": 0.2067,
+      "step": 10414
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.000195280199252802,
+      "loss": 0.5556,
+      "step": 10415
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019526774595267745,
+      "loss": 0.2199,
+      "step": 10416
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001952552926525529,
+      "loss": 0.2447,
+      "step": 10417
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019524283935242837,
+      "loss": 0.1987,
+      "step": 10418
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019523038605230385,
+      "loss": 0.1785,
+      "step": 10419
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001952179327521793,
+      "loss": 0.1431,
+      "step": 10420
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001952054794520548,
+      "loss": 0.3172,
+      "step": 10421
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00019519302615193023,
+      "loss": 0.2277,
+      "step": 10422
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019518057285180572,
+      "loss": 0.2482,
+      "step": 10423
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019516811955168118,
+      "loss": 0.3068,
+      "step": 10424
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019515566625155663,
+      "loss": 0.4882,
+      "step": 10425
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019514321295143212,
+      "loss": 0.3826,
+      "step": 10426
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019513075965130755,
+      "loss": 0.2065,
+      "step": 10427
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019511830635118304,
+      "loss": 0.2612,
+      "step": 10428
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001951058530510585,
+      "loss": 0.2855,
+      "step": 10429
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019509339975093398,
+      "loss": 0.2836,
+      "step": 10430
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019508094645080944,
+      "loss": 0.2292,
+      "step": 10431
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019506849315068493,
+      "loss": 0.1495,
+      "step": 10432
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019505603985056036,
+      "loss": 0.2335,
+      "step": 10433
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019504358655043585,
+      "loss": 0.4878,
+      "step": 10434
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001950311332503113,
+      "loss": 0.2349,
+      "step": 10435
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001950186799501868,
+      "loss": 0.2907,
+      "step": 10436
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019500622665006225,
+      "loss": 0.1996,
+      "step": 10437
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019499377334993774,
+      "loss": 0.1987,
+      "step": 10438
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019498132004981317,
+      "loss": 0.1968,
+      "step": 10439
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019496886674968863,
+      "loss": 0.1661,
+      "step": 10440
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019495641344956412,
+      "loss": 0.2235,
+      "step": 10441
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019494396014943958,
+      "loss": 0.2123,
+      "step": 10442
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019493150684931506,
+      "loss": 0.2789,
+      "step": 10443
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001949190535491905,
+      "loss": 0.2307,
+      "step": 10444
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019490660024906598,
+      "loss": 0.2029,
+      "step": 10445
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019489414694894144,
+      "loss": 0.2204,
+      "step": 10446
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019488169364881693,
+      "loss": 0.1337,
+      "step": 10447
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019486924034869239,
+      "loss": 0.1924,
+      "step": 10448
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00019485678704856787,
+      "loss": 0.3378,
+      "step": 10449
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0001948443337484433,
+      "loss": 0.1792,
+      "step": 10450
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0001948318804483188,
+      "loss": 0.3178,
+      "step": 10451
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019481942714819425,
+      "loss": 0.2395,
+      "step": 10452
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019480697384806974,
+      "loss": 0.1323,
+      "step": 10453
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0001947945205479452,
+      "loss": 0.2539,
+      "step": 10454
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019478206724782063,
+      "loss": 0.2272,
+      "step": 10455
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0001947696139476961,
+      "loss": 0.3491,
+      "step": 10456
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019475716064757157,
+      "loss": 0.1832,
+      "step": 10457
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019474470734744706,
+      "loss": 0.1712,
+      "step": 10458
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019473225404732252,
+      "loss": 0.164,
+      "step": 10459
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.000194719800747198,
+      "loss": 0.2515,
+      "step": 10460
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019470734744707344,
+      "loss": 0.3234,
+      "step": 10461
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019469489414694892,
+      "loss": 0.1951,
+      "step": 10462
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019468244084682438,
+      "loss": 0.2135,
+      "step": 10463
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019466998754669987,
+      "loss": 0.2848,
+      "step": 10464
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019465753424657533,
+      "loss": 0.1245,
+      "step": 10465
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0001946450809464508,
+      "loss": 0.2582,
+      "step": 10466
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019463262764632625,
+      "loss": 0.3141,
+      "step": 10467
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019462017434620173,
+      "loss": 0.346,
+      "step": 10468
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0001946077210460772,
+      "loss": 0.2263,
+      "step": 10469
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019459526774595268,
+      "loss": 0.3081,
+      "step": 10470
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019458281444582814,
+      "loss": 0.2021,
+      "step": 10471
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019457036114570357,
+      "loss": 0.1894,
+      "step": 10472
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019455790784557905,
+      "loss": 0.243,
+      "step": 10473
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019454545454545451,
+      "loss": 0.3359,
+      "step": 10474
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00019453300124533,
+      "loss": 0.2259,
+      "step": 10475
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019452054794520546,
+      "loss": 0.1817,
+      "step": 10476
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019450809464508095,
+      "loss": 0.2149,
+      "step": 10477
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019449564134495638,
+      "loss": 0.2429,
+      "step": 10478
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019448318804483186,
+      "loss": 0.1614,
+      "step": 10479
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019447073474470732,
+      "loss": 0.2209,
+      "step": 10480
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001944582814445828,
+      "loss": 0.2186,
+      "step": 10481
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019444582814445827,
+      "loss": 0.1935,
+      "step": 10482
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019443337484433375,
+      "loss": 0.2342,
+      "step": 10483
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001944209215442092,
+      "loss": 0.1286,
+      "step": 10484
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019440846824408467,
+      "loss": 0.1451,
+      "step": 10485
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019439601494396013,
+      "loss": 0.2635,
+      "step": 10486
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001943835616438356,
+      "loss": 0.2092,
+      "step": 10487
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019437110834371108,
+      "loss": 0.2467,
+      "step": 10488
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001943586550435865,
+      "loss": 0.1616,
+      "step": 10489
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.000194346201743462,
+      "loss": 0.1447,
+      "step": 10490
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019433374844333746,
+      "loss": 0.2401,
+      "step": 10491
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019432129514321294,
+      "loss": 0.1703,
+      "step": 10492
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001943088418430884,
+      "loss": 0.3007,
+      "step": 10493
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001942963885429639,
+      "loss": 0.2494,
+      "step": 10494
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019428393524283932,
+      "loss": 0.1884,
+      "step": 10495
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001942714819427148,
+      "loss": 0.3495,
+      "step": 10496
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019425902864259026,
+      "loss": 0.3006,
+      "step": 10497
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019424657534246575,
+      "loss": 0.2733,
+      "step": 10498
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001942341220423412,
+      "loss": 0.2477,
+      "step": 10499
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001942216687422167,
+      "loss": 0.1757,
+      "step": 10500
+    },
+    {
+      "epoch": 4.02,
+      "eval_cer": 0.11015471057611644,
+      "eval_loss": 0.33645984530448914,
+      "eval_runtime": 47.0368,
+      "eval_samples_per_second": 37.269,
+      "eval_steps_per_second": 4.677,
+      "eval_wer": 0.192469566004435,
+      "step": 10500
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00019420921544209213,
+      "loss": 0.2182,
+      "step": 10501
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0001941967621419676,
+      "loss": 0.2055,
+      "step": 10502
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019418430884184307,
+      "loss": 0.2434,
+      "step": 10503
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019417185554171853,
+      "loss": 0.1822,
+      "step": 10504
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019415940224159402,
+      "loss": 0.3814,
+      "step": 10505
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019414694894146945,
+      "loss": 0.2248,
+      "step": 10506
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019413449564134494,
+      "loss": 0.1615,
+      "step": 10507
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0001941220423412204,
+      "loss": 0.1797,
+      "step": 10508
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019410958904109588,
+      "loss": 0.1708,
+      "step": 10509
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019409713574097134,
+      "loss": 0.1879,
+      "step": 10510
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019408468244084683,
+      "loss": 0.2439,
+      "step": 10511
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019407222914072226,
+      "loss": 0.3192,
+      "step": 10512
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019405977584059775,
+      "loss": 0.2685,
+      "step": 10513
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0001940473225404732,
+      "loss": 0.2289,
+      "step": 10514
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0001940348692403487,
+      "loss": 0.2056,
+      "step": 10515
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019402241594022415,
+      "loss": 0.1115,
+      "step": 10516
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019400996264009964,
+      "loss": 0.2034,
+      "step": 10517
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019399750933997507,
+      "loss": 0.2558,
+      "step": 10518
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019398505603985053,
+      "loss": 0.2634,
+      "step": 10519
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019397260273972601,
+      "loss": 0.1581,
+      "step": 10520
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019396014943960147,
+      "loss": 0.212,
+      "step": 10521
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019394769613947696,
+      "loss": 0.1827,
+      "step": 10522
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0001939352428393524,
+      "loss": 0.2042,
+      "step": 10523
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019392278953922788,
+      "loss": 0.3231,
+      "step": 10524
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019391033623910334,
+      "loss": 0.3497,
+      "step": 10525
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019389788293897882,
+      "loss": 0.1699,
+      "step": 10526
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00019388542963885428,
+      "loss": 0.1773,
+      "step": 10527
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019387297633872977,
+      "loss": 0.1433,
+      "step": 10528
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001938605230386052,
+      "loss": 0.159,
+      "step": 10529
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001938480697384807,
+      "loss": 0.2909,
+      "step": 10530
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019383561643835615,
+      "loss": 0.1766,
+      "step": 10531
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019382316313823163,
+      "loss": 0.179,
+      "step": 10532
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001938107098381071,
+      "loss": 0.1772,
+      "step": 10533
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019379825653798252,
+      "loss": 0.1999,
+      "step": 10534
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.000193785803237858,
+      "loss": 0.1007,
+      "step": 10535
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019377334993773347,
+      "loss": 0.4616,
+      "step": 10536
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019376089663760896,
+      "loss": 0.2522,
+      "step": 10537
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019374844333748442,
+      "loss": 0.2203,
+      "step": 10538
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001937359900373599,
+      "loss": 0.2458,
+      "step": 10539
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019372353673723533,
+      "loss": 0.2083,
+      "step": 10540
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019371108343711082,
+      "loss": 0.1546,
+      "step": 10541
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019369863013698628,
+      "loss": 0.1323,
+      "step": 10542
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019368617683686177,
+      "loss": 0.2546,
+      "step": 10543
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019367372353673722,
+      "loss": 0.2171,
+      "step": 10544
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001936612702366127,
+      "loss": 0.3367,
+      "step": 10545
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019364881693648814,
+      "loss": 0.1503,
+      "step": 10546
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019363636363636363,
+      "loss": 0.2592,
+      "step": 10547
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001936239103362391,
+      "loss": 0.1941,
+      "step": 10548
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019361145703611455,
+      "loss": 0.283,
+      "step": 10549
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019359900373599003,
+      "loss": 0.2153,
+      "step": 10550
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019358655043586547,
+      "loss": 0.1805,
+      "step": 10551
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00019357409713574095,
+      "loss": 0.2495,
+      "step": 10552
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0001935616438356164,
+      "loss": 0.2653,
+      "step": 10553
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001935491905354919,
+      "loss": 0.2819,
+      "step": 10554
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019353673723536736,
+      "loss": 0.473,
+      "step": 10555
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019352428393524284,
+      "loss": 0.273,
+      "step": 10556
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019351183063511828,
+      "loss": 0.1576,
+      "step": 10557
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019349937733499376,
+      "loss": 0.1921,
+      "step": 10558
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019348692403486922,
+      "loss": 0.2616,
+      "step": 10559
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001934744707347447,
+      "loss": 0.1853,
+      "step": 10560
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019346201743462017,
+      "loss": 0.3199,
+      "step": 10561
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019344956413449563,
+      "loss": 0.3068,
+      "step": 10562
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019343711083437108,
+      "loss": 0.2111,
+      "step": 10563
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019342465753424657,
+      "loss": 0.2686,
+      "step": 10564
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019341220423412203,
+      "loss": 0.2951,
+      "step": 10565
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001933997509339975,
+      "loss": 0.1831,
+      "step": 10566
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019338729763387298,
+      "loss": 0.3345,
+      "step": 10567
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001933748443337484,
+      "loss": 0.2292,
+      "step": 10568
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001933623910336239,
+      "loss": 0.225,
+      "step": 10569
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019334993773349935,
+      "loss": 0.1817,
+      "step": 10570
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019333748443337484,
+      "loss": 0.1308,
+      "step": 10571
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001933250311332503,
+      "loss": 0.1856,
+      "step": 10572
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019331257783312576,
+      "loss": 0.1486,
+      "step": 10573
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019330012453300122,
+      "loss": 0.4032,
+      "step": 10574
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.0001932876712328767,
+      "loss": 0.1958,
+      "step": 10575
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019327521793275216,
+      "loss": 0.2293,
+      "step": 10576
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019326276463262765,
+      "loss": 0.1541,
+      "step": 10577
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019325031133250308,
+      "loss": 0.1906,
+      "step": 10578
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00019323785803237857,
+      "loss": 0.171,
+      "step": 10579
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019322540473225403,
+      "loss": 0.5283,
+      "step": 10580
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019321295143212948,
+      "loss": 0.2174,
+      "step": 10581
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019320049813200497,
+      "loss": 0.1945,
+      "step": 10582
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019318804483188043,
+      "loss": 0.1845,
+      "step": 10583
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001931755915317559,
+      "loss": 0.198,
+      "step": 10584
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019316313823163135,
+      "loss": 0.1788,
+      "step": 10585
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019315068493150683,
+      "loss": 0.3021,
+      "step": 10586
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001931382316313823,
+      "loss": 0.2681,
+      "step": 10587
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019312577833125778,
+      "loss": 0.283,
+      "step": 10588
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001931133250311332,
+      "loss": 0.19,
+      "step": 10589
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001931008717310087,
+      "loss": 0.1644,
+      "step": 10590
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019308841843088416,
+      "loss": 0.1715,
+      "step": 10591
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019307596513075964,
+      "loss": 0.2239,
+      "step": 10592
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001930635118306351,
+      "loss": 0.4803,
+      "step": 10593
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001930510585305106,
+      "loss": 0.2895,
+      "step": 10594
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019303860523038602,
+      "loss": 0.1734,
+      "step": 10595
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001930261519302615,
+      "loss": 0.2037,
+      "step": 10596
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019301369863013697,
+      "loss": 0.2292,
+      "step": 10597
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019300124533001243,
+      "loss": 0.2938,
+      "step": 10598
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001929887920298879,
+      "loss": 0.3892,
+      "step": 10599
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019297633872976334,
+      "loss": 0.234,
+      "step": 10600
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019296388542963883,
+      "loss": 0.1855,
+      "step": 10601
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001929514321295143,
+      "loss": 0.1923,
+      "step": 10602
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019293897882938978,
+      "loss": 0.1399,
+      "step": 10603
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019292652552926524,
+      "loss": 0.145,
+      "step": 10604
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00019291407222914072,
+      "loss": 0.3704,
+      "step": 10605
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019290161892901615,
+      "loss": 0.2421,
+      "step": 10606
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019288916562889164,
+      "loss": 0.1815,
+      "step": 10607
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001928767123287671,
+      "loss": 0.2502,
+      "step": 10608
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019286425902864259,
+      "loss": 0.2131,
+      "step": 10609
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019285180572851804,
+      "loss": 0.1709,
+      "step": 10610
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019283935242839353,
+      "loss": 0.3744,
+      "step": 10611
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019282689912826896,
+      "loss": 0.2988,
+      "step": 10612
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019281444582814442,
+      "loss": 0.1824,
+      "step": 10613
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001928019925280199,
+      "loss": 0.1526,
+      "step": 10614
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019278953922789537,
+      "loss": 0.2004,
+      "step": 10615
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019277708592777085,
+      "loss": 0.2043,
+      "step": 10616
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019276463262764629,
+      "loss": 0.2378,
+      "step": 10617
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019275217932752177,
+      "loss": 0.3846,
+      "step": 10618
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019273972602739723,
+      "loss": 0.3063,
+      "step": 10619
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019272727272727272,
+      "loss": 0.1781,
+      "step": 10620
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019271481942714818,
+      "loss": 0.1547,
+      "step": 10621
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019270236612702366,
+      "loss": 0.1141,
+      "step": 10622
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001926899128268991,
+      "loss": 0.2886,
+      "step": 10623
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019267745952677458,
+      "loss": 0.3249,
+      "step": 10624
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019266500622665004,
+      "loss": 0.1993,
+      "step": 10625
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019265255292652553,
+      "loss": 0.2624,
+      "step": 10626
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019264009962640099,
+      "loss": 0.199,
+      "step": 10627
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019262764632627642,
+      "loss": 0.2256,
+      "step": 10628
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0001926151930261519,
+      "loss": 0.1639,
+      "step": 10629
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019260273972602736,
+      "loss": 0.3244,
+      "step": 10630
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00019259028642590285,
+      "loss": 0.285,
+      "step": 10631
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001925778331257783,
+      "loss": 0.4584,
+      "step": 10632
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001925653798256538,
+      "loss": 0.266,
+      "step": 10633
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019255292652552923,
+      "loss": 0.1619,
+      "step": 10634
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019254047322540471,
+      "loss": 0.1512,
+      "step": 10635
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019252801992528017,
+      "loss": 0.3949,
+      "step": 10636
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019251556662515566,
+      "loss": 0.2352,
+      "step": 10637
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019250311332503112,
+      "loss": 0.3006,
+      "step": 10638
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001924906600249066,
+      "loss": 0.2137,
+      "step": 10639
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019247820672478204,
+      "loss": 0.1632,
+      "step": 10640
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019246575342465752,
+      "loss": 0.189,
+      "step": 10641
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019245330012453298,
+      "loss": 0.1431,
+      "step": 10642
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019244084682440847,
+      "loss": 0.3878,
+      "step": 10643
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019242839352428393,
+      "loss": 0.2379,
+      "step": 10644
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019241594022415936,
+      "loss": 0.1912,
+      "step": 10645
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019240348692403485,
+      "loss": 0.2095,
+      "step": 10646
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001923910336239103,
+      "loss": 0.1781,
+      "step": 10647
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001923785803237858,
+      "loss": 0.3193,
+      "step": 10648
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019236612702366125,
+      "loss": 0.3045,
+      "step": 10649
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019235367372353674,
+      "loss": 0.2753,
+      "step": 10650
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019234122042341217,
+      "loss": 0.1745,
+      "step": 10651
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019232876712328766,
+      "loss": 0.2006,
+      "step": 10652
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019231631382316311,
+      "loss": 0.1673,
+      "step": 10653
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0001923038605230386,
+      "loss": 0.154,
+      "step": 10654
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019229140722291406,
+      "loss": 0.3498,
+      "step": 10655
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019227895392278955,
+      "loss": 0.2092,
+      "step": 10656
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00019226650062266498,
+      "loss": 0.2667,
+      "step": 10657
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019225404732254046,
+      "loss": 0.2019,
+      "step": 10658
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019224159402241592,
+      "loss": 0.2467,
+      "step": 10659
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019222914072229138,
+      "loss": 0.2877,
+      "step": 10660
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019221668742216687,
+      "loss": 0.4204,
+      "step": 10661
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001922042341220423,
+      "loss": 0.2461,
+      "step": 10662
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001921917808219178,
+      "loss": 0.2496,
+      "step": 10663
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019217932752179325,
+      "loss": 0.2477,
+      "step": 10664
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019216687422166873,
+      "loss": 0.1487,
+      "step": 10665
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001921544209215442,
+      "loss": 0.2487,
+      "step": 10666
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019214196762141968,
+      "loss": 0.2382,
+      "step": 10667
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001921295143212951,
+      "loss": 0.3257,
+      "step": 10668
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001921170610211706,
+      "loss": 0.2078,
+      "step": 10669
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019210460772104606,
+      "loss": 0.3126,
+      "step": 10670
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019209215442092154,
+      "loss": 0.1683,
+      "step": 10671
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.000192079701120797,
+      "loss": 0.1652,
+      "step": 10672
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001920672478206725,
+      "loss": 0.1737,
+      "step": 10673
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019205479452054792,
+      "loss": 0.469,
+      "step": 10674
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019204234122042338,
+      "loss": 0.2642,
+      "step": 10675
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019202988792029886,
+      "loss": 0.2075,
+      "step": 10676
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019201743462017432,
+      "loss": 0.1605,
+      "step": 10677
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001920049813200498,
+      "loss": 0.2536,
+      "step": 10678
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019199252801992524,
+      "loss": 0.2822,
+      "step": 10679
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019198007471980073,
+      "loss": 0.2533,
+      "step": 10680
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001919676214196762,
+      "loss": 0.1905,
+      "step": 10681
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019195516811955167,
+      "loss": 0.1656,
+      "step": 10682
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00019194271481942713,
+      "loss": 0.1649,
+      "step": 10683
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019193026151930262,
+      "loss": 0.1933,
+      "step": 10684
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019191780821917805,
+      "loss": 0.2779,
+      "step": 10685
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019190535491905354,
+      "loss": 0.2589,
+      "step": 10686
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.000191892901618929,
+      "loss": 0.2806,
+      "step": 10687
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019188044831880448,
+      "loss": 0.1932,
+      "step": 10688
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019186799501867994,
+      "loss": 0.1512,
+      "step": 10689
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019185554171855543,
+      "loss": 0.1252,
+      "step": 10690
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019184308841843086,
+      "loss": 0.1658,
+      "step": 10691
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019183063511830632,
+      "loss": 0.2221,
+      "step": 10692
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001918181818181818,
+      "loss": 0.3983,
+      "step": 10693
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019180572851805727,
+      "loss": 0.2649,
+      "step": 10694
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019179327521793275,
+      "loss": 0.2089,
+      "step": 10695
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019178082191780818,
+      "loss": 0.1928,
+      "step": 10696
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019176836861768367,
+      "loss": 0.3591,
+      "step": 10697
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019175591531755913,
+      "loss": 0.1688,
+      "step": 10698
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019174346201743462,
+      "loss": 0.2403,
+      "step": 10699
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019173100871731007,
+      "loss": 0.2285,
+      "step": 10700
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019171855541718556,
+      "loss": 0.2639,
+      "step": 10701
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.000191706102117061,
+      "loss": 0.1448,
+      "step": 10702
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019169364881693648,
+      "loss": 0.1707,
+      "step": 10703
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019168119551681194,
+      "loss": 0.2282,
+      "step": 10704
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019166874221668742,
+      "loss": 0.2948,
+      "step": 10705
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019165628891656288,
+      "loss": 0.1568,
+      "step": 10706
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019164383561643832,
+      "loss": 0.2282,
+      "step": 10707
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001916313823163138,
+      "loss": 0.1398,
+      "step": 10708
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00019161892901618926,
+      "loss": 0.1572,
+      "step": 10709
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019160647571606475,
+      "loss": 0.1785,
+      "step": 10710
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001915940224159402,
+      "loss": 0.3713,
+      "step": 10711
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001915815691158157,
+      "loss": 0.2613,
+      "step": 10712
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019156911581569113,
+      "loss": 0.1756,
+      "step": 10713
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001915566625155666,
+      "loss": 0.3221,
+      "step": 10714
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019154420921544207,
+      "loss": 0.1153,
+      "step": 10715
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019153175591531756,
+      "loss": 0.1615,
+      "step": 10716
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019151930261519302,
+      "loss": 0.2486,
+      "step": 10717
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001915068493150685,
+      "loss": 0.4734,
+      "step": 10718
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019149439601494393,
+      "loss": 0.3653,
+      "step": 10719
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019148194271481942,
+      "loss": 0.1738,
+      "step": 10720
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019146948941469488,
+      "loss": 0.1596,
+      "step": 10721
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019145703611457034,
+      "loss": 0.2928,
+      "step": 10722
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019144458281444583,
+      "loss": 0.2951,
+      "step": 10723
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019143212951432126,
+      "loss": 0.2967,
+      "step": 10724
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019141967621419674,
+      "loss": 0.1856,
+      "step": 10725
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001914072229140722,
+      "loss": 0.2302,
+      "step": 10726
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001913947696139477,
+      "loss": 0.2285,
+      "step": 10727
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019138231631382315,
+      "loss": 0.1567,
+      "step": 10728
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019136986301369863,
+      "loss": 0.2829,
+      "step": 10729
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019135740971357407,
+      "loss": 0.3142,
+      "step": 10730
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019134495641344955,
+      "loss": 0.2639,
+      "step": 10731
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.000191332503113325,
+      "loss": 0.1906,
+      "step": 10732
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001913200498132005,
+      "loss": 0.1991,
+      "step": 10733
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019130759651307596,
+      "loss": 0.1773,
+      "step": 10734
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019129514321295142,
+      "loss": 0.2628,
+      "step": 10735
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00019128268991282688,
+      "loss": 0.3182,
+      "step": 10736
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019127023661270236,
+      "loss": 0.2173,
+      "step": 10737
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019125778331257782,
+      "loss": 0.2344,
+      "step": 10738
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019124533001245328,
+      "loss": 0.1567,
+      "step": 10739
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019123287671232874,
+      "loss": 0.1747,
+      "step": 10740
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001912204234122042,
+      "loss": 0.1435,
+      "step": 10741
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019120797011207969,
+      "loss": 0.1801,
+      "step": 10742
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019119551681195514,
+      "loss": 0.5834,
+      "step": 10743
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019118306351183063,
+      "loss": 0.2094,
+      "step": 10744
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001911706102117061,
+      "loss": 0.172,
+      "step": 10745
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019115815691158155,
+      "loss": 0.1599,
+      "step": 10746
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.000191145703611457,
+      "loss": 0.199,
+      "step": 10747
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001911332503113325,
+      "loss": 0.2886,
+      "step": 10748
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019112079701120795,
+      "loss": 0.2793,
+      "step": 10749
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019110834371108344,
+      "loss": 0.2018,
+      "step": 10750
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019109589041095887,
+      "loss": 0.2162,
+      "step": 10751
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019108343711083436,
+      "loss": 0.158,
+      "step": 10752
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019107098381070982,
+      "loss": 0.1433,
+      "step": 10753
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019105853051058528,
+      "loss": 0.1394,
+      "step": 10754
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019104607721046076,
+      "loss": 0.4448,
+      "step": 10755
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019103362391033622,
+      "loss": 0.2983,
+      "step": 10756
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019102117061021168,
+      "loss": 0.2173,
+      "step": 10757
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019100871731008714,
+      "loss": 0.1615,
+      "step": 10758
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019099626400996263,
+      "loss": 0.1393,
+      "step": 10759
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019098381070983809,
+      "loss": 0.1632,
+      "step": 10760
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00019097135740971357,
+      "loss": 0.3093,
+      "step": 10761
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.000190958904109589,
+      "loss": 0.2399,
+      "step": 10762
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001909464508094645,
+      "loss": 0.2678,
+      "step": 10763
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019093399750933995,
+      "loss": 0.1749,
+      "step": 10764
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019092154420921544,
+      "loss": 0.1711,
+      "step": 10765
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001909090909090909,
+      "loss": 0.1467,
+      "step": 10766
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019089663760896638,
+      "loss": 0.2168,
+      "step": 10767
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001908841843088418,
+      "loss": 0.3145,
+      "step": 10768
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019087173100871727,
+      "loss": 0.2187,
+      "step": 10769
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019085927770859276,
+      "loss": 0.1685,
+      "step": 10770
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019084682440846822,
+      "loss": 0.1454,
+      "step": 10771
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001908343711083437,
+      "loss": 0.173,
+      "step": 10772
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019082191780821914,
+      "loss": 0.2116,
+      "step": 10773
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019080946450809462,
+      "loss": 0.2805,
+      "step": 10774
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019079701120797008,
+      "loss": 0.2203,
+      "step": 10775
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019078455790784557,
+      "loss": 0.1653,
+      "step": 10776
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019077210460772103,
+      "loss": 0.1535,
+      "step": 10777
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001907596513075965,
+      "loss": 0.2523,
+      "step": 10778
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019074719800747195,
+      "loss": 0.2454,
+      "step": 10779
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019073474470734743,
+      "loss": 0.4051,
+      "step": 10780
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001907222914072229,
+      "loss": 0.1504,
+      "step": 10781
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019070983810709838,
+      "loss": 0.2222,
+      "step": 10782
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019069738480697384,
+      "loss": 0.1432,
+      "step": 10783
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019068493150684932,
+      "loss": 0.1518,
+      "step": 10784
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019067247820672475,
+      "loss": 0.2113,
+      "step": 10785
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019066002490660021,
+      "loss": 0.3343,
+      "step": 10786
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001906475716064757,
+      "loss": 0.2353,
+      "step": 10787
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00019063511830635116,
+      "loss": 0.206,
+      "step": 10788
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019062266500622665,
+      "loss": 0.1833,
+      "step": 10789
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019061021170610208,
+      "loss": 0.188,
+      "step": 10790
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019059775840597756,
+      "loss": 0.2478,
+      "step": 10791
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019058530510585302,
+      "loss": 0.2142,
+      "step": 10792
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001905728518057285,
+      "loss": 0.2453,
+      "step": 10793
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019056039850560397,
+      "loss": 0.2109,
+      "step": 10794
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019054794520547945,
+      "loss": 0.2216,
+      "step": 10795
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001905354919053549,
+      "loss": 0.3091,
+      "step": 10796
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019052303860523037,
+      "loss": 0.1786,
+      "step": 10797
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019051058530510583,
+      "loss": 0.2278,
+      "step": 10798
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019049813200498132,
+      "loss": 0.2594,
+      "step": 10799
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019048567870485678,
+      "loss": 0.3133,
+      "step": 10800
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001904732254047322,
+      "loss": 0.2403,
+      "step": 10801
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001904607721046077,
+      "loss": 0.3209,
+      "step": 10802
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019044831880448316,
+      "loss": 0.2389,
+      "step": 10803
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019043586550435864,
+      "loss": 0.2432,
+      "step": 10804
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001904234122042341,
+      "loss": 0.4054,
+      "step": 10805
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001904109589041096,
+      "loss": 0.2658,
+      "step": 10806
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019039850560398502,
+      "loss": 0.1548,
+      "step": 10807
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001903860523038605,
+      "loss": 0.2197,
+      "step": 10808
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019037359900373596,
+      "loss": 0.1291,
+      "step": 10809
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019036114570361145,
+      "loss": 0.215,
+      "step": 10810
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001903486924034869,
+      "loss": 0.3055,
+      "step": 10811
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001903362391033624,
+      "loss": 0.2183,
+      "step": 10812
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019032378580323783,
+      "loss": 0.2168,
+      "step": 10813
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00019031133250311331,
+      "loss": 0.1728,
+      "step": 10814
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019029887920298877,
+      "loss": 0.1771,
+      "step": 10815
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019028642590286423,
+      "loss": 0.2284,
+      "step": 10816
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019027397260273972,
+      "loss": 0.22,
+      "step": 10817
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019026151930261515,
+      "loss": 0.2876,
+      "step": 10818
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019024906600249064,
+      "loss": 0.194,
+      "step": 10819
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001902366127023661,
+      "loss": 0.2712,
+      "step": 10820
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019022415940224158,
+      "loss": 0.1202,
+      "step": 10821
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019021170610211704,
+      "loss": 0.1841,
+      "step": 10822
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019019925280199253,
+      "loss": 0.1693,
+      "step": 10823
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019018679950186796,
+      "loss": 0.2934,
+      "step": 10824
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019017434620174345,
+      "loss": 0.3527,
+      "step": 10825
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001901618929016189,
+      "loss": 0.2302,
+      "step": 10826
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001901494396014944,
+      "loss": 0.1973,
+      "step": 10827
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019013698630136985,
+      "loss": 0.2001,
+      "step": 10828
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019012453300124534,
+      "loss": 0.2189,
+      "step": 10829
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019011207970112077,
+      "loss": 0.2528,
+      "step": 10830
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019009962640099626,
+      "loss": 0.171,
+      "step": 10831
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019008717310087171,
+      "loss": 0.1556,
+      "step": 10832
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019007471980074717,
+      "loss": 0.2204,
+      "step": 10833
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019006226650062266,
+      "loss": 0.1355,
+      "step": 10834
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001900498132004981,
+      "loss": 0.1548,
+      "step": 10835
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019003735990037358,
+      "loss": 0.3622,
+      "step": 10836
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019002490660024904,
+      "loss": 0.2363,
+      "step": 10837
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00019001245330012452,
+      "loss": 0.1704,
+      "step": 10838
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00018999999999999998,
+      "loss": 0.1502,
+      "step": 10839
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00018998754669987547,
+      "loss": 0.2286,
+      "step": 10840
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001899750933997509,
+      "loss": 0.2453,
+      "step": 10841
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001899626400996264,
+      "loss": 0.1464,
+      "step": 10842
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018995018679950185,
+      "loss": 0.3,
+      "step": 10843
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018993773349937733,
+      "loss": 0.176,
+      "step": 10844
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001899252801992528,
+      "loss": 0.1537,
+      "step": 10845
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018991282689912828,
+      "loss": 0.1982,
+      "step": 10846
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001899003735990037,
+      "loss": 0.1311,
+      "step": 10847
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018988792029887917,
+      "loss": 0.2775,
+      "step": 10848
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018987546699875466,
+      "loss": 0.292,
+      "step": 10849
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018986301369863012,
+      "loss": 0.2919,
+      "step": 10850
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001898505603985056,
+      "loss": 0.2104,
+      "step": 10851
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018983810709838103,
+      "loss": 0.1954,
+      "step": 10852
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018982565379825652,
+      "loss": 0.2107,
+      "step": 10853
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018981320049813198,
+      "loss": 0.1881,
+      "step": 10854
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018980074719800747,
+      "loss": 0.4393,
+      "step": 10855
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018978829389788292,
+      "loss": 0.1934,
+      "step": 10856
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001897758405977584,
+      "loss": 0.1799,
+      "step": 10857
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018976338729763384,
+      "loss": 0.1808,
+      "step": 10858
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018975093399750933,
+      "loss": 0.1318,
+      "step": 10859
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001897384806973848,
+      "loss": 0.215,
+      "step": 10860
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018972602739726027,
+      "loss": 0.3718,
+      "step": 10861
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018971357409713573,
+      "loss": 0.2562,
+      "step": 10862
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018970112079701122,
+      "loss": 0.201,
+      "step": 10863
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00018968866749688665,
+      "loss": 0.1931,
+      "step": 10864
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001896762141967621,
+      "loss": 0.2524,
+      "step": 10865
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0001896637608966376,
+      "loss": 0.1578,
+      "step": 10866
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018965130759651306,
+      "loss": 0.1818,
+      "step": 10867
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018963885429638854,
+      "loss": 0.48,
+      "step": 10868
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018962640099626398,
+      "loss": 0.3137,
+      "step": 10869
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018961394769613946,
+      "loss": 0.2483,
+      "step": 10870
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018960149439601492,
+      "loss": 0.1171,
+      "step": 10871
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001895890410958904,
+      "loss": 0.1743,
+      "step": 10872
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018957658779576587,
+      "loss": 0.2328,
+      "step": 10873
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018956413449564135,
+      "loss": 0.2358,
+      "step": 10874
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018955168119551678,
+      "loss": 0.2391,
+      "step": 10875
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018953922789539227,
+      "loss": 0.1638,
+      "step": 10876
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018952677459526773,
+      "loss": 0.1897,
+      "step": 10877
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018951432129514322,
+      "loss": 0.1873,
+      "step": 10878
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018950186799501868,
+      "loss": 0.1577,
+      "step": 10879
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001894894146948941,
+      "loss": 0.3233,
+      "step": 10880
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001894769613947696,
+      "loss": 0.4089,
+      "step": 10881
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018946450809464505,
+      "loss": 0.2008,
+      "step": 10882
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018945205479452054,
+      "loss": 0.2524,
+      "step": 10883
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.000189439601494396,
+      "loss": 0.1292,
+      "step": 10884
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018942714819427148,
+      "loss": 0.1525,
+      "step": 10885
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018941469489414692,
+      "loss": 0.183,
+      "step": 10886
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001894022415940224,
+      "loss": 0.2865,
+      "step": 10887
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018938978829389786,
+      "loss": 0.1824,
+      "step": 10888
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018937733499377335,
+      "loss": 0.2068,
+      "step": 10889
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001893648816936488,
+      "loss": 0.1882,
+      "step": 10890
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0001893524283935243,
+      "loss": 0.1589,
+      "step": 10891
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00018933997509339973,
+      "loss": 0.0991,
+      "step": 10892
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001893275217932752,
+      "loss": 0.327,
+      "step": 10893
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018931506849315067,
+      "loss": 0.2747,
+      "step": 10894
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018930261519302613,
+      "loss": 0.153,
+      "step": 10895
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018929016189290162,
+      "loss": 0.2067,
+      "step": 10896
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018927770859277705,
+      "loss": 0.1858,
+      "step": 10897
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018926525529265254,
+      "loss": 0.1962,
+      "step": 10898
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.000189252801992528,
+      "loss": 0.3496,
+      "step": 10899
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018924034869240348,
+      "loss": 0.1991,
+      "step": 10900
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018922789539227894,
+      "loss": 0.21,
+      "step": 10901
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001892154420921544,
+      "loss": 0.2462,
+      "step": 10902
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018920298879202986,
+      "loss": 0.168,
+      "step": 10903
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018919053549190534,
+      "loss": 0.1778,
+      "step": 10904
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001891780821917808,
+      "loss": 0.5336,
+      "step": 10905
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001891656288916563,
+      "loss": 0.2212,
+      "step": 10906
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018915317559153175,
+      "loss": 0.319,
+      "step": 10907
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001891407222914072,
+      "loss": 0.1753,
+      "step": 10908
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018912826899128267,
+      "loss": 0.1711,
+      "step": 10909
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018911581569115815,
+      "loss": 0.2138,
+      "step": 10910
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0001891033623910336,
+      "loss": 0.2911,
+      "step": 10911
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018909090909090907,
+      "loss": 0.2374,
+      "step": 10912
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018907845579078453,
+      "loss": 0.2272,
+      "step": 10913
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018906600249066,
+      "loss": 0.2335,
+      "step": 10914
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018905354919053548,
+      "loss": 0.1352,
+      "step": 10915
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018904109589041094,
+      "loss": 0.234,
+      "step": 10916
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018902864259028642,
+      "loss": 0.1549,
+      "step": 10917
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00018901618929016188,
+      "loss": 0.3017,
+      "step": 10918
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018900373599003734,
+      "loss": 0.2342,
+      "step": 10919
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001889912826899128,
+      "loss": 0.1596,
+      "step": 10920
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018897882938978829,
+      "loss": 0.1866,
+      "step": 10921
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018896637608966374,
+      "loss": 0.2529,
+      "step": 10922
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018895392278953923,
+      "loss": 0.1935,
+      "step": 10923
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018894146948941466,
+      "loss": 0.3527,
+      "step": 10924
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018892901618929015,
+      "loss": 0.2073,
+      "step": 10925
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001889165628891656,
+      "loss": 0.1566,
+      "step": 10926
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018890410958904107,
+      "loss": 0.1785,
+      "step": 10927
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018889165628891655,
+      "loss": 0.2435,
+      "step": 10928
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.000188879202988792,
+      "loss": 0.1845,
+      "step": 10929
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018886674968866747,
+      "loss": 0.4013,
+      "step": 10930
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018885429638854293,
+      "loss": 0.2329,
+      "step": 10931
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018884184308841842,
+      "loss": 0.21,
+      "step": 10932
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018882938978829388,
+      "loss": 0.2794,
+      "step": 10933
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018881693648816936,
+      "loss": 0.1827,
+      "step": 10934
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001888044831880448,
+      "loss": 0.2625,
+      "step": 10935
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018879202988792028,
+      "loss": 0.3125,
+      "step": 10936
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018877957658779574,
+      "loss": 0.2395,
+      "step": 10937
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018876712328767123,
+      "loss": 0.235,
+      "step": 10938
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018875466998754669,
+      "loss": 0.198,
+      "step": 10939
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018874221668742217,
+      "loss": 0.2142,
+      "step": 10940
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001887297633872976,
+      "loss": 0.1693,
+      "step": 10941
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018871731008717306,
+      "loss": 0.2118,
+      "step": 10942
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00018870485678704855,
+      "loss": 0.2952,
+      "step": 10943
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.000188692403486924,
+      "loss": 0.1796,
+      "step": 10944
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001886799501867995,
+      "loss": 0.1813,
+      "step": 10945
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018866749688667493,
+      "loss": 0.2414,
+      "step": 10946
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018865504358655041,
+      "loss": 0.395,
+      "step": 10947
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018864259028642587,
+      "loss": 0.2692,
+      "step": 10948
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018863013698630136,
+      "loss": 0.3592,
+      "step": 10949
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018861768368617682,
+      "loss": 0.2478,
+      "step": 10950
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001886052303860523,
+      "loss": 0.2079,
+      "step": 10951
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018859277708592774,
+      "loss": 0.1854,
+      "step": 10952
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018858032378580322,
+      "loss": 0.2118,
+      "step": 10953
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018856787048567868,
+      "loss": 0.2012,
+      "step": 10954
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018855541718555417,
+      "loss": 0.2745,
+      "step": 10955
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018854296388542963,
+      "loss": 0.2076,
+      "step": 10956
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018853051058530511,
+      "loss": 0.1838,
+      "step": 10957
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018851805728518055,
+      "loss": 0.173,
+      "step": 10958
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.000188505603985056,
+      "loss": 0.1844,
+      "step": 10959
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001884931506849315,
+      "loss": 0.237,
+      "step": 10960
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018848069738480695,
+      "loss": 0.4098,
+      "step": 10961
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018846824408468244,
+      "loss": 0.2105,
+      "step": 10962
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018845579078455787,
+      "loss": 0.2886,
+      "step": 10963
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018844333748443336,
+      "loss": 0.138,
+      "step": 10964
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018843088418430881,
+      "loss": 0.2037,
+      "step": 10965
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0001884184308841843,
+      "loss": 0.1794,
+      "step": 10966
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018840597758405976,
+      "loss": 0.2508,
+      "step": 10967
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018839352428393525,
+      "loss": 0.391,
+      "step": 10968
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018838107098381068,
+      "loss": 0.2444,
+      "step": 10969
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00018836861768368616,
+      "loss": 0.191,
+      "step": 10970
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018835616438356162,
+      "loss": 0.2202,
+      "step": 10971
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001883437110834371,
+      "loss": 0.1935,
+      "step": 10972
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018833125778331257,
+      "loss": 0.2002,
+      "step": 10973
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.000188318804483188,
+      "loss": 0.3394,
+      "step": 10974
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001883063511830635,
+      "loss": 0.287,
+      "step": 10975
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018829389788293895,
+      "loss": 0.1644,
+      "step": 10976
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018828144458281443,
+      "loss": 0.1785,
+      "step": 10977
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001882689912826899,
+      "loss": 0.1557,
+      "step": 10978
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018825653798256538,
+      "loss": 0.2063,
+      "step": 10979
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001882440846824408,
+      "loss": 0.322,
+      "step": 10980
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001882316313823163,
+      "loss": 0.218,
+      "step": 10981
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018821917808219176,
+      "loss": 0.1799,
+      "step": 10982
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018820672478206724,
+      "loss": 0.1687,
+      "step": 10983
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001881942714819427,
+      "loss": 0.198,
+      "step": 10984
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001881818181818182,
+      "loss": 0.1886,
+      "step": 10985
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018816936488169362,
+      "loss": 0.3198,
+      "step": 10986
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001881569115815691,
+      "loss": 0.1529,
+      "step": 10987
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018814445828144457,
+      "loss": 0.3182,
+      "step": 10988
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018813200498132002,
+      "loss": 0.1759,
+      "step": 10989
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001881195516811955,
+      "loss": 0.1522,
+      "step": 10990
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018810709838107094,
+      "loss": 0.2186,
+      "step": 10991
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018809464508094643,
+      "loss": 0.2555,
+      "step": 10992
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0001880821917808219,
+      "loss": 0.3487,
+      "step": 10993
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018806973848069737,
+      "loss": 0.1747,
+      "step": 10994
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018805728518057283,
+      "loss": 0.2252,
+      "step": 10995
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00018804483188044832,
+      "loss": 0.1599,
+      "step": 10996
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018803237858032375,
+      "loss": 0.2464,
+      "step": 10997
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018801992528019924,
+      "loss": 0.2391,
+      "step": 10998
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001880074719800747,
+      "loss": 0.423,
+      "step": 10999
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018799501867995018,
+      "loss": 0.1881,
+      "step": 11000
+    },
+    {
+      "epoch": 4.22,
+      "eval_cer": 0.10663519129094563,
+      "eval_loss": 0.35278430581092834,
+      "eval_runtime": 47.148,
+      "eval_samples_per_second": 37.181,
+      "eval_steps_per_second": 4.666,
+      "eval_wer": 0.17746752952889533,
+      "step": 11000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018798256537982564,
+      "loss": 0.1673,
+      "step": 11001
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018797011207970113,
+      "loss": 0.2718,
+      "step": 11002
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018795765877957656,
+      "loss": 0.1823,
+      "step": 11003
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018794520547945205,
+      "loss": 0.1407,
+      "step": 11004
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001879327521793275,
+      "loss": 0.4092,
+      "step": 11005
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018792029887920297,
+      "loss": 0.1866,
+      "step": 11006
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018790784557907845,
+      "loss": 0.2497,
+      "step": 11007
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018789539227895388,
+      "loss": 0.2889,
+      "step": 11008
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018788293897882937,
+      "loss": 0.173,
+      "step": 11009
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018787048567870483,
+      "loss": 0.2837,
+      "step": 11010
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018785803237858032,
+      "loss": 0.4441,
+      "step": 11011
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018784557907845577,
+      "loss": 0.2378,
+      "step": 11012
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018783312577833126,
+      "loss": 0.1722,
+      "step": 11013
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001878206724782067,
+      "loss": 0.2878,
+      "step": 11014
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018780821917808218,
+      "loss": 0.1953,
+      "step": 11015
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018779576587795764,
+      "loss": 0.3413,
+      "step": 11016
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018778331257783312,
+      "loss": 0.3127,
+      "step": 11017
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018777085927770858,
+      "loss": 0.6032,
+      "step": 11018
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018775840597758407,
+      "loss": 0.1999,
+      "step": 11019
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001877459526774595,
+      "loss": 0.17,
+      "step": 11020
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018773349937733496,
+      "loss": 0.3565,
+      "step": 11021
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00018772104607721045,
+      "loss": 0.2224,
+      "step": 11022
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001877085927770859,
+      "loss": 0.1993,
+      "step": 11023
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001876961394769614,
+      "loss": 0.3355,
+      "step": 11024
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018768368617683683,
+      "loss": 0.3383,
+      "step": 11025
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001876712328767123,
+      "loss": 0.1615,
+      "step": 11026
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018765877957658777,
+      "loss": 0.1723,
+      "step": 11027
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018764632627646326,
+      "loss": 0.1276,
+      "step": 11028
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018763387297633872,
+      "loss": 0.293,
+      "step": 11029
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001876214196762142,
+      "loss": 0.2751,
+      "step": 11030
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018760896637608963,
+      "loss": 0.2545,
+      "step": 11031
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018759651307596512,
+      "loss": 0.1926,
+      "step": 11032
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018758405977584058,
+      "loss": 0.1302,
+      "step": 11033
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018757160647571607,
+      "loss": 0.1548,
+      "step": 11034
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018755915317559153,
+      "loss": 0.2205,
+      "step": 11035
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018754669987546696,
+      "loss": 0.2859,
+      "step": 11036
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018753424657534244,
+      "loss": 0.2524,
+      "step": 11037
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001875217932752179,
+      "loss": 0.2684,
+      "step": 11038
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001875093399750934,
+      "loss": 0.2992,
+      "step": 11039
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018749688667496885,
+      "loss": 0.1634,
+      "step": 11040
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018748443337484433,
+      "loss": 0.2316,
+      "step": 11041
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018747198007471977,
+      "loss": 0.1649,
+      "step": 11042
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018745952677459525,
+      "loss": 0.2941,
+      "step": 11043
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001874470734744707,
+      "loss": 0.2253,
+      "step": 11044
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001874346201743462,
+      "loss": 0.2545,
+      "step": 11045
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018742216687422166,
+      "loss": 0.3178,
+      "step": 11046
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018740971357409714,
+      "loss": 0.1556,
+      "step": 11047
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018739726027397258,
+      "loss": 0.1831,
+      "step": 11048
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00018738480697384806,
+      "loss": 0.2602,
+      "step": 11049
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018737235367372352,
+      "loss": 0.2009,
+      "step": 11050
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.000187359900373599,
+      "loss": 0.2271,
+      "step": 11051
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018734744707347447,
+      "loss": 0.2518,
+      "step": 11052
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001873349937733499,
+      "loss": 0.151,
+      "step": 11053
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018732254047322539,
+      "loss": 0.1951,
+      "step": 11054
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018731008717310084,
+      "loss": 0.4065,
+      "step": 11055
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018729763387297633,
+      "loss": 0.2443,
+      "step": 11056
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001872851805728518,
+      "loss": 0.2496,
+      "step": 11057
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018727272727272728,
+      "loss": 0.2041,
+      "step": 11058
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001872602739726027,
+      "loss": 0.2239,
+      "step": 11059
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001872478206724782,
+      "loss": 0.1921,
+      "step": 11060
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018723536737235365,
+      "loss": 0.4264,
+      "step": 11061
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018722291407222914,
+      "loss": 0.2896,
+      "step": 11062
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001872104607721046,
+      "loss": 0.189,
+      "step": 11063
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018719800747198006,
+      "loss": 0.2057,
+      "step": 11064
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018718555417185552,
+      "loss": 0.241,
+      "step": 11065
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.000187173100871731,
+      "loss": 0.1525,
+      "step": 11066
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018716064757160646,
+      "loss": 0.2554,
+      "step": 11067
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018714819427148192,
+      "loss": 0.3033,
+      "step": 11068
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001871357409713574,
+      "loss": 0.2617,
+      "step": 11069
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018712328767123284,
+      "loss": 0.239,
+      "step": 11070
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018711083437110833,
+      "loss": 0.2709,
+      "step": 11071
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018709838107098379,
+      "loss": 0.1466,
+      "step": 11072
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018708592777085927,
+      "loss": 0.1408,
+      "step": 11073
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00018707347447073473,
+      "loss": 0.3514,
+      "step": 11074
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001870610211706102,
+      "loss": 0.3262,
+      "step": 11075
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018704856787048565,
+      "loss": 0.1472,
+      "step": 11076
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018703611457036114,
+      "loss": 0.1729,
+      "step": 11077
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001870236612702366,
+      "loss": 0.2564,
+      "step": 11078
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018701120797011208,
+      "loss": 0.2402,
+      "step": 11079
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018699875466998754,
+      "loss": 0.2584,
+      "step": 11080
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.000186986301369863,
+      "loss": 0.1954,
+      "step": 11081
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018697384806973846,
+      "loss": 0.5491,
+      "step": 11082
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018696139476961392,
+      "loss": 0.164,
+      "step": 11083
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001869489414694894,
+      "loss": 0.221,
+      "step": 11084
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018693648816936486,
+      "loss": 0.2235,
+      "step": 11085
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018692403486924032,
+      "loss": 0.3369,
+      "step": 11086
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018691158156911578,
+      "loss": 0.2915,
+      "step": 11087
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018689912826899127,
+      "loss": 0.1689,
+      "step": 11088
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018688667496886673,
+      "loss": 0.2856,
+      "step": 11089
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001868742216687422,
+      "loss": 0.1424,
+      "step": 11090
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018686176836861767,
+      "loss": 0.2073,
+      "step": 11091
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018684931506849313,
+      "loss": 0.1488,
+      "step": 11092
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001868368617683686,
+      "loss": 0.3533,
+      "step": 11093
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018682440846824408,
+      "loss": 0.1482,
+      "step": 11094
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018681195516811954,
+      "loss": 0.1886,
+      "step": 11095
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018679950186799502,
+      "loss": 0.1843,
+      "step": 11096
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018678704856787045,
+      "loss": 0.2869,
+      "step": 11097
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018677459526774594,
+      "loss": 0.2247,
+      "step": 11098
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001867621419676214,
+      "loss": 0.4618,
+      "step": 11099
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018674968866749686,
+      "loss": 0.1511,
+      "step": 11100
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00018673723536737235,
+      "loss": 0.1991,
+      "step": 11101
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001867247820672478,
+      "loss": 0.145,
+      "step": 11102
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018671232876712326,
+      "loss": 0.1709,
+      "step": 11103
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018669987546699872,
+      "loss": 0.3792,
+      "step": 11104
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001866874221668742,
+      "loss": 0.2812,
+      "step": 11105
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018667496886674967,
+      "loss": 0.2255,
+      "step": 11106
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018666251556662515,
+      "loss": 0.1952,
+      "step": 11107
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001866500622665006,
+      "loss": 0.1828,
+      "step": 11108
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018663760896637607,
+      "loss": 0.3113,
+      "step": 11109
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018662515566625153,
+      "loss": 0.2618,
+      "step": 11110
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018661270236612702,
+      "loss": 0.3711,
+      "step": 11111
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018660024906600248,
+      "loss": 0.2069,
+      "step": 11112
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018658779576587796,
+      "loss": 0.1972,
+      "step": 11113
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001865753424657534,
+      "loss": 0.1714,
+      "step": 11114
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018656288916562886,
+      "loss": 0.2654,
+      "step": 11115
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018655043586550434,
+      "loss": 0.2044,
+      "step": 11116
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001865379825653798,
+      "loss": 0.2372,
+      "step": 11117
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001865255292652553,
+      "loss": 0.3254,
+      "step": 11118
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018651307596513072,
+      "loss": 0.1723,
+      "step": 11119
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001865006226650062,
+      "loss": 0.2506,
+      "step": 11120
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018648816936488166,
+      "loss": 0.2142,
+      "step": 11121
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018647571606475715,
+      "loss": 0.2228,
+      "step": 11122
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001864632627646326,
+      "loss": 0.1415,
+      "step": 11123
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001864508094645081,
+      "loss": 0.206,
+      "step": 11124
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018643835616438353,
+      "loss": 0.1866,
+      "step": 11125
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018642590286425901,
+      "loss": 0.2049,
+      "step": 11126
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00018641344956413447,
+      "loss": 0.2104,
+      "step": 11127
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018640099626400996,
+      "loss": 0.1694,
+      "step": 11128
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018638854296388542,
+      "loss": 0.1343,
+      "step": 11129
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001863760896637609,
+      "loss": 0.3937,
+      "step": 11130
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018636363636363634,
+      "loss": 0.237,
+      "step": 11131
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001863511830635118,
+      "loss": 0.1942,
+      "step": 11132
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018633872976338728,
+      "loss": 0.2035,
+      "step": 11133
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018632627646326274,
+      "loss": 0.1434,
+      "step": 11134
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018631382316313823,
+      "loss": 0.1572,
+      "step": 11135
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018630136986301366,
+      "loss": 0.3956,
+      "step": 11136
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018628891656288915,
+      "loss": 0.2343,
+      "step": 11137
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001862764632627646,
+      "loss": 0.1782,
+      "step": 11138
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001862640099626401,
+      "loss": 0.2418,
+      "step": 11139
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018625155666251555,
+      "loss": 0.2333,
+      "step": 11140
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018623910336239104,
+      "loss": 0.193,
+      "step": 11141
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018622665006226647,
+      "loss": 0.2475,
+      "step": 11142
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018621419676214196,
+      "loss": 0.2646,
+      "step": 11143
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018620174346201742,
+      "loss": 0.1755,
+      "step": 11144
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001861892901618929,
+      "loss": 0.2964,
+      "step": 11145
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018617683686176836,
+      "loss": 0.3225,
+      "step": 11146
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001861643835616438,
+      "loss": 0.348,
+      "step": 11147
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018615193026151928,
+      "loss": 0.1446,
+      "step": 11148
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018613947696139474,
+      "loss": 0.4034,
+      "step": 11149
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018612702366127022,
+      "loss": 0.1521,
+      "step": 11150
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018611457036114568,
+      "loss": 0.2234,
+      "step": 11151
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00018610211706102117,
+      "loss": 0.1942,
+      "step": 11152
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001860896637608966,
+      "loss": 0.2819,
+      "step": 11153
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001860772104607721,
+      "loss": 0.1811,
+      "step": 11154
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018606475716064755,
+      "loss": 0.2373,
+      "step": 11155
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018605230386052303,
+      "loss": 0.1916,
+      "step": 11156
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001860398505603985,
+      "loss": 0.3473,
+      "step": 11157
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018602739726027398,
+      "loss": 0.151,
+      "step": 11158
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001860149439601494,
+      "loss": 0.1709,
+      "step": 11159
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001860024906600249,
+      "loss": 0.2196,
+      "step": 11160
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018599003735990036,
+      "loss": 0.5491,
+      "step": 11161
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018597758405977582,
+      "loss": 0.2214,
+      "step": 11162
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001859651307596513,
+      "loss": 0.1777,
+      "step": 11163
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018595267745952673,
+      "loss": 0.1529,
+      "step": 11164
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018594022415940222,
+      "loss": 0.2421,
+      "step": 11165
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018592777085927768,
+      "loss": 0.1778,
+      "step": 11166
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018591531755915317,
+      "loss": 0.2689,
+      "step": 11167
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018590286425902862,
+      "loss": 0.3061,
+      "step": 11168
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001858904109589041,
+      "loss": 0.1907,
+      "step": 11169
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018587795765877954,
+      "loss": 0.2667,
+      "step": 11170
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018586550435865503,
+      "loss": 0.1914,
+      "step": 11171
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001858530510585305,
+      "loss": 0.1321,
+      "step": 11172
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018584059775840597,
+      "loss": 0.3398,
+      "step": 11173
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018582814445828143,
+      "loss": 0.2928,
+      "step": 11174
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018581569115815692,
+      "loss": 0.2181,
+      "step": 11175
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018580323785803235,
+      "loss": 0.2076,
+      "step": 11176
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018579078455790784,
+      "loss": 0.2105,
+      "step": 11177
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0001857783312577833,
+      "loss": 0.1638,
+      "step": 11178
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00018576587795765876,
+      "loss": 0.1969,
+      "step": 11179
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018575342465753424,
+      "loss": 0.4434,
+      "step": 11180
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018574097135740968,
+      "loss": 0.1498,
+      "step": 11181
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018572851805728516,
+      "loss": 0.3106,
+      "step": 11182
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018571606475716062,
+      "loss": 0.1412,
+      "step": 11183
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0001857036114570361,
+      "loss": 0.1884,
+      "step": 11184
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018569115815691157,
+      "loss": 0.1893,
+      "step": 11185
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018567870485678705,
+      "loss": 0.3025,
+      "step": 11186
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018566625155666248,
+      "loss": 0.3421,
+      "step": 11187
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018565379825653797,
+      "loss": 0.1963,
+      "step": 11188
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018564134495641343,
+      "loss": 0.1978,
+      "step": 11189
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018562889165628892,
+      "loss": 0.2241,
+      "step": 11190
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018561643835616438,
+      "loss": 0.1616,
+      "step": 11191
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018560398505603986,
+      "loss": 0.313,
+      "step": 11192
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0001855915317559153,
+      "loss": 0.3264,
+      "step": 11193
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018557907845579075,
+      "loss": 0.1846,
+      "step": 11194
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018556662515566624,
+      "loss": 0.2486,
+      "step": 11195
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0001855541718555417,
+      "loss": 0.2453,
+      "step": 11196
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018554171855541718,
+      "loss": 0.2295,
+      "step": 11197
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018552926525529262,
+      "loss": 0.1689,
+      "step": 11198
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0001855168119551681,
+      "loss": 0.5303,
+      "step": 11199
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018550435865504356,
+      "loss": 0.2374,
+      "step": 11200
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018549190535491905,
+      "loss": 0.2817,
+      "step": 11201
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0001854794520547945,
+      "loss": 0.1954,
+      "step": 11202
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018546699875467,
+      "loss": 0.1643,
+      "step": 11203
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00018545454545454543,
+      "loss": 0.1631,
+      "step": 11204
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0001854420921544209,
+      "loss": 0.2515,
+      "step": 11205
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018542963885429637,
+      "loss": 0.268,
+      "step": 11206
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018541718555417186,
+      "loss": 0.212,
+      "step": 11207
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018540473225404732,
+      "loss": 0.1792,
+      "step": 11208
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018539227895392275,
+      "loss": 0.1238,
+      "step": 11209
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018537982565379824,
+      "loss": 0.1708,
+      "step": 11210
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0001853673723536737,
+      "loss": 0.3885,
+      "step": 11211
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018535491905354918,
+      "loss": 0.216,
+      "step": 11212
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018534246575342464,
+      "loss": 0.3013,
+      "step": 11213
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018533001245330013,
+      "loss": 0.2103,
+      "step": 11214
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018531755915317556,
+      "loss": 0.1357,
+      "step": 11215
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018530510585305104,
+      "loss": 0.1545,
+      "step": 11216
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0001852926525529265,
+      "loss": 0.2652,
+      "step": 11217
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.000185280199252802,
+      "loss": 0.3131,
+      "step": 11218
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018526774595267745,
+      "loss": 0.1548,
+      "step": 11219
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018525529265255294,
+      "loss": 0.2346,
+      "step": 11220
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018524283935242837,
+      "loss": 0.1866,
+      "step": 11221
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018523038605230385,
+      "loss": 0.278,
+      "step": 11222
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0001852179327521793,
+      "loss": 0.1425,
+      "step": 11223
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0001852054794520548,
+      "loss": 0.2859,
+      "step": 11224
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018519302615193026,
+      "loss": 0.2101,
+      "step": 11225
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.0001851805728518057,
+      "loss": 0.2104,
+      "step": 11226
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018516811955168118,
+      "loss": 0.1622,
+      "step": 11227
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018515566625155664,
+      "loss": 0.2031,
+      "step": 11228
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018514321295143212,
+      "loss": 0.1927,
+      "step": 11229
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018513075965130758,
+      "loss": 0.4702,
+      "step": 11230
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00018511830635118307,
+      "loss": 0.2186,
+      "step": 11231
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0001851058530510585,
+      "loss": 0.174,
+      "step": 11232
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018509339975093399,
+      "loss": 0.1456,
+      "step": 11233
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018508094645080944,
+      "loss": 0.1669,
+      "step": 11234
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018506849315068493,
+      "loss": 0.2971,
+      "step": 11235
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0001850560398505604,
+      "loss": 0.2596,
+      "step": 11236
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018504358655043585,
+      "loss": 0.2421,
+      "step": 11237
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0001850311332503113,
+      "loss": 0.2063,
+      "step": 11238
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0001850186799501868,
+      "loss": 0.268,
+      "step": 11239
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018500622665006225,
+      "loss": 0.22,
+      "step": 11240
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0001849937733499377,
+      "loss": 0.2147,
+      "step": 11241
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0001849813200498132,
+      "loss": 0.2039,
+      "step": 11242
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018496886674968863,
+      "loss": 0.386,
+      "step": 11243
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018495641344956412,
+      "loss": 0.3071,
+      "step": 11244
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018494396014943958,
+      "loss": 0.2184,
+      "step": 11245
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018493150684931506,
+      "loss": 0.1826,
+      "step": 11246
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018491905354919052,
+      "loss": 0.1844,
+      "step": 11247
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018490660024906598,
+      "loss": 0.1684,
+      "step": 11248
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018489414694894144,
+      "loss": 0.3751,
+      "step": 11249
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018488169364881693,
+      "loss": 0.1686,
+      "step": 11250
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018486924034869239,
+      "loss": 0.1948,
+      "step": 11251
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018485678704856787,
+      "loss": 0.2619,
+      "step": 11252
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018484433374844333,
+      "loss": 0.1867,
+      "step": 11253
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0001848318804483188,
+      "loss": 0.1815,
+      "step": 11254
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00018481942714819425,
+      "loss": 0.2885,
+      "step": 11255
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0001848069738480697,
+      "loss": 0.1782,
+      "step": 11256
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0001847945205479452,
+      "loss": 0.2744,
+      "step": 11257
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018478206724782065,
+      "loss": 0.2293,
+      "step": 11258
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018476961394769611,
+      "loss": 0.214,
+      "step": 11259
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018475716064757157,
+      "loss": 0.2423,
+      "step": 11260
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018474470734744706,
+      "loss": 0.3715,
+      "step": 11261
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018473225404732252,
+      "loss": 0.3315,
+      "step": 11262
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.000184719800747198,
+      "loss": 0.3964,
+      "step": 11263
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018470734744707346,
+      "loss": 0.3563,
+      "step": 11264
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018469489414694892,
+      "loss": 0.1685,
+      "step": 11265
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018468244084682438,
+      "loss": 0.1864,
+      "step": 11266
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018466998754669987,
+      "loss": 0.1872,
+      "step": 11267
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018465753424657533,
+      "loss": 0.316,
+      "step": 11268
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018464508094645081,
+      "loss": 0.2522,
+      "step": 11269
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018463262764632625,
+      "loss": 0.2259,
+      "step": 11270
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018462017434620173,
+      "loss": 0.2407,
+      "step": 11271
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0001846077210460772,
+      "loss": 0.4098,
+      "step": 11272
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018459526774595265,
+      "loss": 0.2142,
+      "step": 11273
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018458281444582814,
+      "loss": 0.2082,
+      "step": 11274
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018457036114570357,
+      "loss": 0.2849,
+      "step": 11275
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018455790784557906,
+      "loss": 0.1556,
+      "step": 11276
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018454545454545451,
+      "loss": 0.1482,
+      "step": 11277
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018453300124533,
+      "loss": 0.1546,
+      "step": 11278
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018452054794520546,
+      "loss": 0.2633,
+      "step": 11279
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018450809464508095,
+      "loss": 0.266,
+      "step": 11280
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018449564134495638,
+      "loss": 0.2162,
+      "step": 11281
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018448318804483186,
+      "loss": 0.3599,
+      "step": 11282
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00018447073474470732,
+      "loss": 0.1926,
+      "step": 11283
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0001844582814445828,
+      "loss": 0.2568,
+      "step": 11284
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018444582814445827,
+      "loss": 0.1693,
+      "step": 11285
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018443337484433376,
+      "loss": 0.2741,
+      "step": 11286
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0001844209215442092,
+      "loss": 0.2374,
+      "step": 11287
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018440846824408465,
+      "loss": 0.2272,
+      "step": 11288
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018439601494396013,
+      "loss": 0.216,
+      "step": 11289
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0001843835616438356,
+      "loss": 0.209,
+      "step": 11290
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018437110834371108,
+      "loss": 0.1939,
+      "step": 11291
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0001843586550435865,
+      "loss": 0.2723,
+      "step": 11292
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.000184346201743462,
+      "loss": 0.3286,
+      "step": 11293
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018433374844333746,
+      "loss": 0.1976,
+      "step": 11294
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018432129514321294,
+      "loss": 0.1824,
+      "step": 11295
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0001843088418430884,
+      "loss": 0.224,
+      "step": 11296
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0001842963885429639,
+      "loss": 0.2616,
+      "step": 11297
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018428393524283932,
+      "loss": 0.3435,
+      "step": 11298
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0001842714819427148,
+      "loss": 0.2851,
+      "step": 11299
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018425902864259027,
+      "loss": 0.2187,
+      "step": 11300
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018424657534246575,
+      "loss": 0.2469,
+      "step": 11301
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0001842341220423412,
+      "loss": 0.1912,
+      "step": 11302
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018422166874221664,
+      "loss": 0.2227,
+      "step": 11303
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018420921544209213,
+      "loss": 0.1826,
+      "step": 11304
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.0001841967621419676,
+      "loss": 0.2482,
+      "step": 11305
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018418430884184307,
+      "loss": 0.3515,
+      "step": 11306
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018417185554171853,
+      "loss": 0.2465,
+      "step": 11307
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018415940224159402,
+      "loss": 0.1579,
+      "step": 11308
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018414694894146945,
+      "loss": 0.1832,
+      "step": 11309
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00018413449564134494,
+      "loss": 0.2586,
+      "step": 11310
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0001841220423412204,
+      "loss": 0.356,
+      "step": 11311
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018410958904109588,
+      "loss": 0.2561,
+      "step": 11312
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018409713574097134,
+      "loss": 0.1842,
+      "step": 11313
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018408468244084683,
+      "loss": 0.2316,
+      "step": 11314
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018407222914072226,
+      "loss": 0.2109,
+      "step": 11315
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018405977584059775,
+      "loss": 0.2162,
+      "step": 11316
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0001840473225404732,
+      "loss": 0.2078,
+      "step": 11317
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0001840348692403487,
+      "loss": 0.3122,
+      "step": 11318
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018402241594022415,
+      "loss": 0.2425,
+      "step": 11319
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018400996264009958,
+      "loss": 0.1659,
+      "step": 11320
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018399750933997507,
+      "loss": 0.1426,
+      "step": 11321
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018398505603985053,
+      "loss": 0.1567,
+      "step": 11322
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018397260273972602,
+      "loss": 0.1758,
+      "step": 11323
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018396014943960147,
+      "loss": 0.2897,
+      "step": 11324
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018394769613947696,
+      "loss": 0.3588,
+      "step": 11325
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0001839352428393524,
+      "loss": 0.3929,
+      "step": 11326
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018392278953922788,
+      "loss": 0.1695,
+      "step": 11327
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018391033623910334,
+      "loss": 0.2505,
+      "step": 11328
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018389788293897882,
+      "loss": 0.2413,
+      "step": 11329
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018388542963885428,
+      "loss": 0.2951,
+      "step": 11330
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018387297633872977,
+      "loss": 0.2915,
+      "step": 11331
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0001838605230386052,
+      "loss": 0.1907,
+      "step": 11332
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0001838480697384807,
+      "loss": 0.1547,
+      "step": 11333
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00018383561643835615,
+      "loss": 0.18,
+      "step": 11334
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0001838231631382316,
+      "loss": 0.2571,
+      "step": 11335
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0001838107098381071,
+      "loss": 0.3797,
+      "step": 11336
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018379825653798253,
+      "loss": 0.1871,
+      "step": 11337
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.000183785803237858,
+      "loss": 0.3082,
+      "step": 11338
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018377334993773347,
+      "loss": 0.2212,
+      "step": 11339
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018376089663760896,
+      "loss": 0.2401,
+      "step": 11340
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018374844333748442,
+      "loss": 0.2093,
+      "step": 11341
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0001837359900373599,
+      "loss": 0.1541,
+      "step": 11342
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018372353673723533,
+      "loss": 0.3955,
+      "step": 11343
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018371108343711082,
+      "loss": 0.1839,
+      "step": 11344
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018369863013698628,
+      "loss": 0.2842,
+      "step": 11345
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018368617683686177,
+      "loss": 0.2052,
+      "step": 11346
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018367372353673723,
+      "loss": 0.3134,
+      "step": 11347
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0001836612702366127,
+      "loss": 0.202,
+      "step": 11348
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018364881693648814,
+      "loss": 0.2937,
+      "step": 11349
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0001836363636363636,
+      "loss": 0.3344,
+      "step": 11350
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0001836239103362391,
+      "loss": 0.1928,
+      "step": 11351
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018361145703611455,
+      "loss": 0.227,
+      "step": 11352
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018359900373599003,
+      "loss": 0.1923,
+      "step": 11353
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018358655043586547,
+      "loss": 0.2367,
+      "step": 11354
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018357409713574095,
+      "loss": 0.37,
+      "step": 11355
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0001835616438356164,
+      "loss": 0.1995,
+      "step": 11356
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0001835491905354919,
+      "loss": 0.3276,
+      "step": 11357
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018353673723536736,
+      "loss": 0.1796,
+      "step": 11358
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018352428393524284,
+      "loss": 0.249,
+      "step": 11359
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018351183063511828,
+      "loss": 0.2189,
+      "step": 11360
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018349937733499376,
+      "loss": 0.2445,
+      "step": 11361
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00018348692403486922,
+      "loss": 0.2305,
+      "step": 11362
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001834744707347447,
+      "loss": 0.2062,
+      "step": 11363
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018346201743462017,
+      "loss": 0.2366,
+      "step": 11364
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018344956413449565,
+      "loss": 0.1616,
+      "step": 11365
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018343711083437109,
+      "loss": 0.2064,
+      "step": 11366
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018342465753424654,
+      "loss": 0.2558,
+      "step": 11367
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018341220423412203,
+      "loss": 0.3606,
+      "step": 11368
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001833997509339975,
+      "loss": 0.2043,
+      "step": 11369
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018338729763387298,
+      "loss": 0.2389,
+      "step": 11370
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001833748443337484,
+      "loss": 0.2122,
+      "step": 11371
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001833623910336239,
+      "loss": 0.1802,
+      "step": 11372
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018334993773349935,
+      "loss": 0.1899,
+      "step": 11373
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018333748443337484,
+      "loss": 0.3529,
+      "step": 11374
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001833250311332503,
+      "loss": 0.3025,
+      "step": 11375
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018331257783312579,
+      "loss": 0.2026,
+      "step": 11376
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018330012453300122,
+      "loss": 0.2075,
+      "step": 11377
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001832876712328767,
+      "loss": 0.144,
+      "step": 11378
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018327521793275216,
+      "loss": 0.1586,
+      "step": 11379
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018326276463262765,
+      "loss": 0.345,
+      "step": 11380
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001832503113325031,
+      "loss": 0.2628,
+      "step": 11381
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018323785803237854,
+      "loss": 0.2136,
+      "step": 11382
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018322540473225403,
+      "loss": 0.176,
+      "step": 11383
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018321295143212949,
+      "loss": 0.293,
+      "step": 11384
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018320049813200497,
+      "loss": 0.1941,
+      "step": 11385
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018318804483188043,
+      "loss": 0.3722,
+      "step": 11386
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018317559153175592,
+      "loss": 0.2217,
+      "step": 11387
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00018316313823163135,
+      "loss": 0.1362,
+      "step": 11388
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018315068493150684,
+      "loss": 0.1771,
+      "step": 11389
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001831382316313823,
+      "loss": 0.2477,
+      "step": 11390
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018312577833125778,
+      "loss": 0.1326,
+      "step": 11391
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018311332503113324,
+      "loss": 0.2679,
+      "step": 11392
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018310087173100873,
+      "loss": 0.4565,
+      "step": 11393
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018308841843088416,
+      "loss": 0.1724,
+      "step": 11394
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018307596513075965,
+      "loss": 0.1395,
+      "step": 11395
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001830635118306351,
+      "loss": 0.1306,
+      "step": 11396
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001830510585305106,
+      "loss": 0.1767,
+      "step": 11397
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018303860523038605,
+      "loss": 0.1807,
+      "step": 11398
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018302615193026148,
+      "loss": 0.3368,
+      "step": 11399
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018301369863013697,
+      "loss": 0.1944,
+      "step": 11400
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018300124533001243,
+      "loss": 0.2089,
+      "step": 11401
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001829887920298879,
+      "loss": 0.2209,
+      "step": 11402
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018297633872976337,
+      "loss": 0.2943,
+      "step": 11403
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018296388542963886,
+      "loss": 0.4444,
+      "step": 11404
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001829514321295143,
+      "loss": 0.2137,
+      "step": 11405
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018293897882938978,
+      "loss": 0.2118,
+      "step": 11406
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018292652552926524,
+      "loss": 0.1682,
+      "step": 11407
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018291407222914072,
+      "loss": 0.2051,
+      "step": 11408
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018290161892901618,
+      "loss": 0.1375,
+      "step": 11409
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018288916562889164,
+      "loss": 0.1916,
+      "step": 11410
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001828767123287671,
+      "loss": 0.3873,
+      "step": 11411
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001828642590286426,
+      "loss": 0.1749,
+      "step": 11412
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00018285180572851805,
+      "loss": 0.2219,
+      "step": 11413
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001828393524283935,
+      "loss": 0.157,
+      "step": 11414
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.000182826899128269,
+      "loss": 0.1748,
+      "step": 11415
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018281444582814442,
+      "loss": 0.1462,
+      "step": 11416
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001828019925280199,
+      "loss": 0.2205,
+      "step": 11417
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018278953922789537,
+      "loss": 0.2438,
+      "step": 11418
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018277708592777085,
+      "loss": 0.1659,
+      "step": 11419
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018276463262764631,
+      "loss": 0.1546,
+      "step": 11420
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018275217932752177,
+      "loss": 0.2238,
+      "step": 11421
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018273972602739723,
+      "loss": 0.3478,
+      "step": 11422
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018272727272727272,
+      "loss": 0.2934,
+      "step": 11423
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018271481942714818,
+      "loss": 0.3387,
+      "step": 11424
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018270236612702366,
+      "loss": 0.2087,
+      "step": 11425
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001826899128268991,
+      "loss": 0.1842,
+      "step": 11426
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018267745952677458,
+      "loss": 0.2381,
+      "step": 11427
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018266500622665004,
+      "loss": 0.1524,
+      "step": 11428
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001826525529265255,
+      "loss": 0.2413,
+      "step": 11429
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.000182640099626401,
+      "loss": 0.369,
+      "step": 11430
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018262764632627645,
+      "loss": 0.2085,
+      "step": 11431
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001826151930261519,
+      "loss": 0.1935,
+      "step": 11432
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018260273972602736,
+      "loss": 0.2692,
+      "step": 11433
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018259028642590285,
+      "loss": 0.1445,
+      "step": 11434
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001825778331257783,
+      "loss": 0.1553,
+      "step": 11435
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001825653798256538,
+      "loss": 0.2898,
+      "step": 11436
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018255292652552923,
+      "loss": 0.438,
+      "step": 11437
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018254047322540471,
+      "loss": 0.2045,
+      "step": 11438
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018252801992528017,
+      "loss": 0.205,
+      "step": 11439
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00018251556662515566,
+      "loss": 0.2994,
+      "step": 11440
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018250311332503112,
+      "loss": 0.288,
+      "step": 11441
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0001824906600249066,
+      "loss": 0.2282,
+      "step": 11442
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018247820672478204,
+      "loss": 0.2464,
+      "step": 11443
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018246575342465752,
+      "loss": 0.2057,
+      "step": 11444
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018245330012453298,
+      "loss": 0.2179,
+      "step": 11445
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018244084682440844,
+      "loss": 0.1752,
+      "step": 11446
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018242839352428393,
+      "loss": 0.2752,
+      "step": 11447
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018241594022415936,
+      "loss": 0.2552,
+      "step": 11448
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018240348692403485,
+      "loss": 0.3899,
+      "step": 11449
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0001823910336239103,
+      "loss": 0.2731,
+      "step": 11450
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0001823785803237858,
+      "loss": 0.1748,
+      "step": 11451
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018236612702366125,
+      "loss": 0.2143,
+      "step": 11452
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018235367372353674,
+      "loss": 0.2644,
+      "step": 11453
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018234122042341217,
+      "loss": 0.354,
+      "step": 11454
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018232876712328766,
+      "loss": 0.401,
+      "step": 11455
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018231631382316312,
+      "loss": 0.2265,
+      "step": 11456
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0001823038605230386,
+      "loss": 0.1535,
+      "step": 11457
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018229140722291406,
+      "loss": 0.1709,
+      "step": 11458
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018227895392278955,
+      "loss": 0.1703,
+      "step": 11459
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018226650062266498,
+      "loss": 0.2158,
+      "step": 11460
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018225404732254044,
+      "loss": 0.4281,
+      "step": 11461
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018224159402241592,
+      "loss": 0.3002,
+      "step": 11462
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018222914072229138,
+      "loss": 0.1882,
+      "step": 11463
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00018221668742216687,
+      "loss": 0.207,
+      "step": 11464
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0001822042341220423,
+      "loss": 0.1663,
+      "step": 11465
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0001821917808219178,
+      "loss": 0.1916,
+      "step": 11466
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018217932752179325,
+      "loss": 0.2921,
+      "step": 11467
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018216687422166873,
+      "loss": 0.3416,
+      "step": 11468
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0001821544209215442,
+      "loss": 0.1868,
+      "step": 11469
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018214196762141968,
+      "loss": 0.2004,
+      "step": 11470
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0001821295143212951,
+      "loss": 0.2898,
+      "step": 11471
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0001821170610211706,
+      "loss": 0.179,
+      "step": 11472
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018210460772104606,
+      "loss": 0.3593,
+      "step": 11473
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018209215442092154,
+      "loss": 0.2619,
+      "step": 11474
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.000182079701120797,
+      "loss": 0.1781,
+      "step": 11475
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018206724782067243,
+      "loss": 0.247,
+      "step": 11476
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018205479452054792,
+      "loss": 0.15,
+      "step": 11477
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018204234122042338,
+      "loss": 0.2303,
+      "step": 11478
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018202988792029887,
+      "loss": 0.216,
+      "step": 11479
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018201743462017432,
+      "loss": 0.2855,
+      "step": 11480
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0001820049813200498,
+      "loss": 0.2084,
+      "step": 11481
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018199252801992524,
+      "loss": 0.2143,
+      "step": 11482
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018198007471980073,
+      "loss": 0.214,
+      "step": 11483
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0001819676214196762,
+      "loss": 0.1912,
+      "step": 11484
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018195516811955168,
+      "loss": 0.1881,
+      "step": 11485
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018194271481942713,
+      "loss": 0.4408,
+      "step": 11486
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018193026151930262,
+      "loss": 0.2042,
+      "step": 11487
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018191780821917805,
+      "loss": 0.2576,
+      "step": 11488
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018190535491905354,
+      "loss": 0.206,
+      "step": 11489
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.000181892901618929,
+      "loss": 0.1796,
+      "step": 11490
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018188044831880448,
+      "loss": 0.1717,
+      "step": 11491
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00018186799501867994,
+      "loss": 0.2003,
+      "step": 11492
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018185554171855538,
+      "loss": 0.3408,
+      "step": 11493
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018184308841843086,
+      "loss": 0.1883,
+      "step": 11494
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018183063511830632,
+      "loss": 0.1783,
+      "step": 11495
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0001818181818181818,
+      "loss": 0.2276,
+      "step": 11496
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018180572851805727,
+      "loss": 0.205,
+      "step": 11497
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018179327521793275,
+      "loss": 0.2296,
+      "step": 11498
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018178082191780818,
+      "loss": 0.3385,
+      "step": 11499
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018176836861768367,
+      "loss": 0.3106,
+      "step": 11500
+    },
+    {
+      "epoch": 4.41,
+      "eval_cer": 0.10627621593148252,
+      "eval_loss": 0.39093655347824097,
+      "eval_runtime": 46.5723,
+      "eval_samples_per_second": 37.64,
+      "eval_steps_per_second": 4.724,
+      "eval_wer": 0.1754310539892293,
+      "step": 11500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018175591531755913,
+      "loss": 0.249,
+      "step": 11501
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018174346201743462,
+      "loss": 0.1791,
+      "step": 11502
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018173100871731008,
+      "loss": 0.2056,
+      "step": 11503
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018171855541718556,
+      "loss": 0.1768,
+      "step": 11504
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.000181706102117061,
+      "loss": 0.2866,
+      "step": 11505
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018169364881693648,
+      "loss": 0.4114,
+      "step": 11506
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018168119551681194,
+      "loss": 0.1612,
+      "step": 11507
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0001816687422166874,
+      "loss": 0.1419,
+      "step": 11508
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018165628891656288,
+      "loss": 0.1701,
+      "step": 11509
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018164383561643832,
+      "loss": 0.2683,
+      "step": 11510
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0001816313823163138,
+      "loss": 0.4565,
+      "step": 11511
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018161892901618926,
+      "loss": 0.2001,
+      "step": 11512
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018160647571606475,
+      "loss": 0.2334,
+      "step": 11513
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0001815940224159402,
+      "loss": 0.2552,
+      "step": 11514
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0001815815691158157,
+      "loss": 0.1438,
+      "step": 11515
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018156911581569113,
+      "loss": 0.1636,
+      "step": 11516
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0001815566625155666,
+      "loss": 0.2414,
+      "step": 11517
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00018154420921544207,
+      "loss": 0.3652,
+      "step": 11518
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018153175591531756,
+      "loss": 0.1723,
+      "step": 11519
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018151930261519302,
+      "loss": 0.1719,
+      "step": 11520
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0001815068493150685,
+      "loss": 0.2142,
+      "step": 11521
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018149439601494394,
+      "loss": 0.1527,
+      "step": 11522
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0001814819427148194,
+      "loss": 0.2011,
+      "step": 11523
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018146948941469488,
+      "loss": 0.371,
+      "step": 11524
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018145703611457034,
+      "loss": 0.2223,
+      "step": 11525
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018144458281444583,
+      "loss": 0.1817,
+      "step": 11526
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018143212951432126,
+      "loss": 0.2245,
+      "step": 11527
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018141967621419674,
+      "loss": 0.1499,
+      "step": 11528
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0001814072229140722,
+      "loss": 0.2561,
+      "step": 11529
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0001813947696139477,
+      "loss": 0.3074,
+      "step": 11530
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018138231631382315,
+      "loss": 0.3714,
+      "step": 11531
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018136986301369864,
+      "loss": 0.1509,
+      "step": 11532
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018135740971357407,
+      "loss": 0.199,
+      "step": 11533
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018134495641344955,
+      "loss": 0.2474,
+      "step": 11534
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.000181332503113325,
+      "loss": 0.2648,
+      "step": 11535
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0001813200498132005,
+      "loss": 0.2634,
+      "step": 11536
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018130759651307596,
+      "loss": 0.199,
+      "step": 11537
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018129514321295144,
+      "loss": 0.1733,
+      "step": 11538
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018128268991282688,
+      "loss": 0.4591,
+      "step": 11539
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018127023661270234,
+      "loss": 0.2243,
+      "step": 11540
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018125778331257782,
+      "loss": 0.1347,
+      "step": 11541
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018124533001245328,
+      "loss": 0.2779,
+      "step": 11542
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00018123287671232877,
+      "loss": 0.2568,
+      "step": 11543
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0001812204234122042,
+      "loss": 0.342,
+      "step": 11544
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018120797011207969,
+      "loss": 0.1801,
+      "step": 11545
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018119551681195515,
+      "loss": 0.1969,
+      "step": 11546
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018118306351183063,
+      "loss": 0.2312,
+      "step": 11547
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0001811706102117061,
+      "loss": 0.1385,
+      "step": 11548
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018115815691158158,
+      "loss": 0.405,
+      "step": 11549
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.000181145703611457,
+      "loss": 0.2779,
+      "step": 11550
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0001811332503113325,
+      "loss": 0.1679,
+      "step": 11551
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018112079701120795,
+      "loss": 0.3113,
+      "step": 11552
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018110834371108344,
+      "loss": 0.1977,
+      "step": 11553
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0001810958904109589,
+      "loss": 0.2651,
+      "step": 11554
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018108343711083433,
+      "loss": 0.4103,
+      "step": 11555
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018107098381070982,
+      "loss": 0.2183,
+      "step": 11556
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018105853051058528,
+      "loss": 0.2434,
+      "step": 11557
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018104607721046076,
+      "loss": 0.2457,
+      "step": 11558
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018103362391033622,
+      "loss": 0.2358,
+      "step": 11559
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0001810211706102117,
+      "loss": 0.1771,
+      "step": 11560
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018100871731008714,
+      "loss": 0.3615,
+      "step": 11561
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018099626400996263,
+      "loss": 0.2485,
+      "step": 11562
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0001809838107098381,
+      "loss": 0.2463,
+      "step": 11563
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018097135740971357,
+      "loss": 0.2894,
+      "step": 11564
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018095890410958903,
+      "loss": 0.3477,
+      "step": 11565
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018094645080946452,
+      "loss": 0.2187,
+      "step": 11566
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018093399750933995,
+      "loss": 0.2669,
+      "step": 11567
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018092154420921544,
+      "loss": 0.2742,
+      "step": 11568
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0001809090909090909,
+      "loss": 0.2062,
+      "step": 11569
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00018089663760896635,
+      "loss": 0.1795,
+      "step": 11570
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018088418430884184,
+      "loss": 0.3082,
+      "step": 11571
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018087173100871727,
+      "loss": 0.2493,
+      "step": 11572
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018085927770859276,
+      "loss": 0.3044,
+      "step": 11573
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018084682440846822,
+      "loss": 0.3213,
+      "step": 11574
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0001808343711083437,
+      "loss": 0.2096,
+      "step": 11575
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018082191780821916,
+      "loss": 0.2916,
+      "step": 11576
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018080946450809465,
+      "loss": 0.1964,
+      "step": 11577
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018079701120797008,
+      "loss": 0.2849,
+      "step": 11578
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018078455790784557,
+      "loss": 0.2959,
+      "step": 11579
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018077210460772103,
+      "loss": 0.3553,
+      "step": 11580
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018075965130759651,
+      "loss": 0.2409,
+      "step": 11581
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018074719800747197,
+      "loss": 0.2122,
+      "step": 11582
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018073474470734743,
+      "loss": 0.1435,
+      "step": 11583
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0001807222914072229,
+      "loss": 0.2446,
+      "step": 11584
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018070983810709838,
+      "loss": 0.1467,
+      "step": 11585
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018069738480697384,
+      "loss": 0.3134,
+      "step": 11586
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0001806849315068493,
+      "loss": 0.2325,
+      "step": 11587
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018067247820672476,
+      "loss": 0.2608,
+      "step": 11588
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018066002490660021,
+      "loss": 0.1449,
+      "step": 11589
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0001806475716064757,
+      "loss": 0.1936,
+      "step": 11590
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018063511830635116,
+      "loss": 0.2511,
+      "step": 11591
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018062266500622665,
+      "loss": 0.2808,
+      "step": 11592
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0001806102117061021,
+      "loss": 0.3543,
+      "step": 11593
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018059775840597756,
+      "loss": 0.2109,
+      "step": 11594
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018058530510585302,
+      "loss": 0.165,
+      "step": 11595
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0001805728518057285,
+      "loss": 0.2775,
+      "step": 11596
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00018056039850560397,
+      "loss": 0.3602,
+      "step": 11597
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018054794520547946,
+      "loss": 0.1842,
+      "step": 11598
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0001805354919053549,
+      "loss": 0.3824,
+      "step": 11599
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018052303860523037,
+      "loss": 0.315,
+      "step": 11600
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018051058530510583,
+      "loss": 0.1885,
+      "step": 11601
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0001804981320049813,
+      "loss": 0.2191,
+      "step": 11602
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018048567870485678,
+      "loss": 0.244,
+      "step": 11603
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018047322540473224,
+      "loss": 0.2779,
+      "step": 11604
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0001804607721046077,
+      "loss": 0.4356,
+      "step": 11605
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018044831880448316,
+      "loss": 0.2339,
+      "step": 11606
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018043586550435864,
+      "loss": 0.2027,
+      "step": 11607
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0001804234122042341,
+      "loss": 0.2304,
+      "step": 11608
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0001804109589041096,
+      "loss": 0.1727,
+      "step": 11609
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018039850560398502,
+      "loss": 0.1804,
+      "step": 11610
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0001803860523038605,
+      "loss": 0.4803,
+      "step": 11611
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018037359900373597,
+      "loss": 0.2647,
+      "step": 11612
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018036114570361145,
+      "loss": 0.3096,
+      "step": 11613
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0001803486924034869,
+      "loss": 0.2329,
+      "step": 11614
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0001803362391033624,
+      "loss": 0.2189,
+      "step": 11615
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018032378580323783,
+      "loss": 0.1976,
+      "step": 11616
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0001803113325031133,
+      "loss": 0.2911,
+      "step": 11617
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018029887920298877,
+      "loss": 0.2897,
+      "step": 11618
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018028642590286423,
+      "loss": 0.1935,
+      "step": 11619
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018027397260273972,
+      "loss": 0.1458,
+      "step": 11620
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018026151930261515,
+      "loss": 0.2225,
+      "step": 11621
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00018024906600249064,
+      "loss": 0.2477,
+      "step": 11622
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0001802366127023661,
+      "loss": 0.19,
+      "step": 11623
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018022415940224158,
+      "loss": 0.3757,
+      "step": 11624
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018021170610211704,
+      "loss": 0.1648,
+      "step": 11625
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018019925280199253,
+      "loss": 0.1505,
+      "step": 11626
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018018679950186796,
+      "loss": 0.184,
+      "step": 11627
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018017434620174345,
+      "loss": 0.1914,
+      "step": 11628
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0001801618929016189,
+      "loss": 0.1414,
+      "step": 11629
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0001801494396014944,
+      "loss": 0.5129,
+      "step": 11630
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018013698630136985,
+      "loss": 0.2154,
+      "step": 11631
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018012453300124534,
+      "loss": 0.1302,
+      "step": 11632
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018011207970112077,
+      "loss": 0.21,
+      "step": 11633
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018009962640099623,
+      "loss": 0.2888,
+      "step": 11634
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018008717310087172,
+      "loss": 0.1601,
+      "step": 11635
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018007471980074718,
+      "loss": 0.2532,
+      "step": 11636
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018006226650062266,
+      "loss": 0.2654,
+      "step": 11637
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0001800498132004981,
+      "loss": 0.3412,
+      "step": 11638
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018003735990037358,
+      "loss": 0.22,
+      "step": 11639
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018002490660024904,
+      "loss": 0.1771,
+      "step": 11640
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00018001245330012453,
+      "loss": 0.1583,
+      "step": 11641
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 0.3631,
+      "step": 11642
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00017998754669987547,
+      "loss": 0.3857,
+      "step": 11643
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0001799750933997509,
+      "loss": 0.1839,
+      "step": 11644
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0001799626400996264,
+      "loss": 0.2066,
+      "step": 11645
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00017995018679950185,
+      "loss": 0.283,
+      "step": 11646
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00017993773349937733,
+      "loss": 0.2979,
+      "step": 11647
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0001799252801992528,
+      "loss": 0.163,
+      "step": 11648
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00017991282689912823,
+      "loss": 0.2717,
+      "step": 11649
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0001799003735990037,
+      "loss": 0.2745,
+      "step": 11650
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017988792029887917,
+      "loss": 0.2528,
+      "step": 11651
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017987546699875466,
+      "loss": 0.1663,
+      "step": 11652
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017986301369863012,
+      "loss": 0.2266,
+      "step": 11653
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0001798505603985056,
+      "loss": 0.28,
+      "step": 11654
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017983810709838103,
+      "loss": 0.3336,
+      "step": 11655
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017982565379825652,
+      "loss": 0.1968,
+      "step": 11656
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017981320049813198,
+      "loss": 0.1798,
+      "step": 11657
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017980074719800747,
+      "loss": 0.2714,
+      "step": 11658
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017978829389788293,
+      "loss": 0.1644,
+      "step": 11659
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0001797758405977584,
+      "loss": 0.2771,
+      "step": 11660
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017976338729763384,
+      "loss": 0.3073,
+      "step": 11661
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017975093399750933,
+      "loss": 0.1972,
+      "step": 11662
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0001797384806973848,
+      "loss": 0.3548,
+      "step": 11663
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017972602739726025,
+      "loss": 0.2463,
+      "step": 11664
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017971357409713573,
+      "loss": 0.1709,
+      "step": 11665
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017970112079701117,
+      "loss": 0.2025,
+      "step": 11666
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017968866749688665,
+      "loss": 0.2165,
+      "step": 11667
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0001796762141967621,
+      "loss": 0.3545,
+      "step": 11668
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0001796637608966376,
+      "loss": 0.2199,
+      "step": 11669
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017965130759651306,
+      "loss": 0.1765,
+      "step": 11670
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017963885429638854,
+      "loss": 0.208,
+      "step": 11671
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017962640099626398,
+      "loss": 0.2027,
+      "step": 11672
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017961394769613946,
+      "loss": 0.236,
+      "step": 11673
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00017960149439601492,
+      "loss": 0.3366,
+      "step": 11674
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0001795890410958904,
+      "loss": 0.1517,
+      "step": 11675
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017957658779576587,
+      "loss": 0.2486,
+      "step": 11676
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017956413449564135,
+      "loss": 0.1792,
+      "step": 11677
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017955168119551679,
+      "loss": 0.1468,
+      "step": 11678
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017953922789539227,
+      "loss": 0.1842,
+      "step": 11679
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017952677459526773,
+      "loss": 0.4094,
+      "step": 11680
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001795143212951432,
+      "loss": 0.1457,
+      "step": 11681
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017950186799501868,
+      "loss": 0.2536,
+      "step": 11682
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001794894146948941,
+      "loss": 0.1918,
+      "step": 11683
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001794769613947696,
+      "loss": 0.2479,
+      "step": 11684
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017946450809464505,
+      "loss": 0.1611,
+      "step": 11685
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017945205479452054,
+      "loss": 0.324,
+      "step": 11686
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.000179439601494396,
+      "loss": 0.3107,
+      "step": 11687
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017942714819427149,
+      "loss": 0.2098,
+      "step": 11688
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017941469489414692,
+      "loss": 0.2232,
+      "step": 11689
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001794022415940224,
+      "loss": 0.1884,
+      "step": 11690
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017938978829389786,
+      "loss": 0.1421,
+      "step": 11691
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017937733499377335,
+      "loss": 0.2619,
+      "step": 11692
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001793648816936488,
+      "loss": 0.364,
+      "step": 11693
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001793524283935243,
+      "loss": 0.249,
+      "step": 11694
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017933997509339973,
+      "loss": 0.145,
+      "step": 11695
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017932752179327519,
+      "loss": 0.3121,
+      "step": 11696
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017931506849315067,
+      "loss": 0.1441,
+      "step": 11697
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017930261519302613,
+      "loss": 0.1728,
+      "step": 11698
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017929016189290162,
+      "loss": 0.2442,
+      "step": 11699
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017927770859277705,
+      "loss": 0.207,
+      "step": 11700
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00017926525529265254,
+      "loss": 0.2607,
+      "step": 11701
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.000179252801992528,
+      "loss": 0.1427,
+      "step": 11702
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017924034869240348,
+      "loss": 0.256,
+      "step": 11703
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017922789539227894,
+      "loss": 0.1477,
+      "step": 11704
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017921544209215443,
+      "loss": 0.2692,
+      "step": 11705
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017920298879202986,
+      "loss": 0.3119,
+      "step": 11706
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017919053549190535,
+      "loss": 0.3696,
+      "step": 11707
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001791780821917808,
+      "loss": 0.2327,
+      "step": 11708
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001791656288916563,
+      "loss": 0.1995,
+      "step": 11709
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017915317559153175,
+      "loss": 0.1946,
+      "step": 11710
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017914072229140724,
+      "loss": 0.3015,
+      "step": 11711
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017912826899128267,
+      "loss": 0.2109,
+      "step": 11712
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017911581569115813,
+      "loss": 0.2993,
+      "step": 11713
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017910336239103361,
+      "loss": 0.1925,
+      "step": 11714
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017909090909090907,
+      "loss": 0.2601,
+      "step": 11715
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017907845579078456,
+      "loss": 0.3026,
+      "step": 11716
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017906600249066,
+      "loss": 0.2512,
+      "step": 11717
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017905354919053548,
+      "loss": 0.3144,
+      "step": 11718
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017904109589041094,
+      "loss": 0.302,
+      "step": 11719
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017902864259028642,
+      "loss": 0.1801,
+      "step": 11720
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017901618929016188,
+      "loss": 0.1847,
+      "step": 11721
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017900373599003737,
+      "loss": 0.2358,
+      "step": 11722
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001789912826899128,
+      "loss": 0.1774,
+      "step": 11723
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001789788293897883,
+      "loss": 0.2325,
+      "step": 11724
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017896637608966375,
+      "loss": 0.1824,
+      "step": 11725
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00017895392278953923,
+      "loss": 0.2332,
+      "step": 11726
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001789414694894147,
+      "loss": 0.2155,
+      "step": 11727
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017892901618929012,
+      "loss": 0.2938,
+      "step": 11728
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001789165628891656,
+      "loss": 0.1447,
+      "step": 11729
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017890410958904107,
+      "loss": 0.5216,
+      "step": 11730
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017889165628891655,
+      "loss": 0.1328,
+      "step": 11731
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017887920298879201,
+      "loss": 0.1709,
+      "step": 11732
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001788667496886675,
+      "loss": 0.3989,
+      "step": 11733
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017885429638854293,
+      "loss": 0.2316,
+      "step": 11734
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017884184308841842,
+      "loss": 0.2662,
+      "step": 11735
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017882938978829388,
+      "loss": 0.3338,
+      "step": 11736
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017881693648816936,
+      "loss": 0.2134,
+      "step": 11737
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017880448318804482,
+      "loss": 0.3152,
+      "step": 11738
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001787920298879203,
+      "loss": 0.1767,
+      "step": 11739
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017877957658779574,
+      "loss": 0.2914,
+      "step": 11740
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017876712328767123,
+      "loss": 0.1791,
+      "step": 11741
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001787546699875467,
+      "loss": 0.2615,
+      "step": 11742
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017874221668742215,
+      "loss": 0.2932,
+      "step": 11743
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017872976338729763,
+      "loss": 0.2812,
+      "step": 11744
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017871731008717306,
+      "loss": 0.2148,
+      "step": 11745
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017870485678704855,
+      "loss": 0.1672,
+      "step": 11746
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.000178692403486924,
+      "loss": 0.155,
+      "step": 11747
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001786799501867995,
+      "loss": 0.1611,
+      "step": 11748
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017866749688667496,
+      "loss": 0.3825,
+      "step": 11749
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017865504358655041,
+      "loss": 0.2485,
+      "step": 11750
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017864259028642587,
+      "loss": 0.1756,
+      "step": 11751
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017863013698630136,
+      "loss": 0.2036,
+      "step": 11752
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00017861768368617682,
+      "loss": 0.2773,
+      "step": 11753
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0001786052303860523,
+      "loss": 0.2074,
+      "step": 11754
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017859277708592776,
+      "loss": 0.5019,
+      "step": 11755
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017858032378580322,
+      "loss": 0.268,
+      "step": 11756
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017856787048567868,
+      "loss": 0.1542,
+      "step": 11757
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017855541718555417,
+      "loss": 0.3039,
+      "step": 11758
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017854296388542963,
+      "loss": 0.1332,
+      "step": 11759
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0001785305105853051,
+      "loss": 0.1885,
+      "step": 11760
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017851805728518055,
+      "loss": 0.2958,
+      "step": 11761
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.000178505603985056,
+      "loss": 0.3286,
+      "step": 11762
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0001784931506849315,
+      "loss": 0.1744,
+      "step": 11763
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017848069738480695,
+      "loss": 0.2176,
+      "step": 11764
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017846824408468244,
+      "loss": 0.2396,
+      "step": 11765
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0001784557907845579,
+      "loss": 0.1712,
+      "step": 11766
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017844333748443336,
+      "loss": 0.2619,
+      "step": 11767
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017843088418430882,
+      "loss": 0.3443,
+      "step": 11768
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0001784184308841843,
+      "loss": 0.2179,
+      "step": 11769
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017840597758405976,
+      "loss": 0.2529,
+      "step": 11770
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017839352428393525,
+      "loss": 0.2125,
+      "step": 11771
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017838107098381068,
+      "loss": 0.1488,
+      "step": 11772
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017836861768368617,
+      "loss": 0.2372,
+      "step": 11773
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017835616438356162,
+      "loss": 0.3294,
+      "step": 11774
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017834371108343708,
+      "loss": 0.2083,
+      "step": 11775
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017833125778331257,
+      "loss": 0.2019,
+      "step": 11776
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017831880448318803,
+      "loss": 0.1864,
+      "step": 11777
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0001783063511830635,
+      "loss": 0.2381,
+      "step": 11778
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00017829389788293895,
+      "loss": 0.2574,
+      "step": 11779
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017828144458281443,
+      "loss": 0.2652,
+      "step": 11780
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0001782689912826899,
+      "loss": 0.2271,
+      "step": 11781
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017825653798256538,
+      "loss": 0.215,
+      "step": 11782
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0001782440846824408,
+      "loss": 0.1593,
+      "step": 11783
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0001782316313823163,
+      "loss": 0.1989,
+      "step": 11784
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017821917808219176,
+      "loss": 0.1121,
+      "step": 11785
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017820672478206724,
+      "loss": 0.3632,
+      "step": 11786
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0001781942714819427,
+      "loss": 0.1959,
+      "step": 11787
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0001781818181818182,
+      "loss": 0.1821,
+      "step": 11788
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017816936488169362,
+      "loss": 0.1406,
+      "step": 11789
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017815691158156908,
+      "loss": 0.2124,
+      "step": 11790
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017814445828144457,
+      "loss": 0.2656,
+      "step": 11791
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017813200498132003,
+      "loss": 0.3406,
+      "step": 11792
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0001781195516811955,
+      "loss": 0.2679,
+      "step": 11793
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017810709838107094,
+      "loss": 0.2726,
+      "step": 11794
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017809464508094643,
+      "loss": 0.2044,
+      "step": 11795
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0001780821917808219,
+      "loss": 0.2355,
+      "step": 11796
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017806973848069738,
+      "loss": 0.1964,
+      "step": 11797
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017805728518057283,
+      "loss": 0.1748,
+      "step": 11798
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017804483188044832,
+      "loss": 0.2574,
+      "step": 11799
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017803237858032375,
+      "loss": 0.1862,
+      "step": 11800
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017801992528019924,
+      "loss": 0.2147,
+      "step": 11801
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0001780074719800747,
+      "loss": 0.2796,
+      "step": 11802
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017799501867995018,
+      "loss": 0.1531,
+      "step": 11803
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017798256537982564,
+      "loss": 0.2006,
+      "step": 11804
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00017797011207970113,
+      "loss": 0.4163,
+      "step": 11805
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017795765877957656,
+      "loss": 0.195,
+      "step": 11806
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017794520547945202,
+      "loss": 0.2131,
+      "step": 11807
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0001779327521793275,
+      "loss": 0.199,
+      "step": 11808
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017792029887920297,
+      "loss": 0.1664,
+      "step": 11809
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017790784557907845,
+      "loss": 0.1876,
+      "step": 11810
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017789539227895388,
+      "loss": 0.3004,
+      "step": 11811
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017788293897882937,
+      "loss": 0.2156,
+      "step": 11812
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017787048567870483,
+      "loss": 0.2248,
+      "step": 11813
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017785803237858032,
+      "loss": 0.2711,
+      "step": 11814
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017784557907845578,
+      "loss": 0.2286,
+      "step": 11815
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017783312577833126,
+      "loss": 0.1748,
+      "step": 11816
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0001778206724782067,
+      "loss": 0.276,
+      "step": 11817
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017780821917808218,
+      "loss": 0.4591,
+      "step": 11818
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017779576587795764,
+      "loss": 0.3042,
+      "step": 11819
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017778331257783313,
+      "loss": 0.1758,
+      "step": 11820
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017777085927770858,
+      "loss": 0.1542,
+      "step": 11821
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017775840597758402,
+      "loss": 0.2002,
+      "step": 11822
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0001777459526774595,
+      "loss": 0.196,
+      "step": 11823
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017773349937733496,
+      "loss": 0.3436,
+      "step": 11824
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017772104607721045,
+      "loss": 0.2181,
+      "step": 11825
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0001777085927770859,
+      "loss": 0.1512,
+      "step": 11826
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0001776961394769614,
+      "loss": 0.1414,
+      "step": 11827
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017768368617683683,
+      "loss": 0.2409,
+      "step": 11828
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0001776712328767123,
+      "loss": 0.2941,
+      "step": 11829
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017765877957658777,
+      "loss": 0.3264,
+      "step": 11830
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00017764632627646326,
+      "loss": 0.2765,
+      "step": 11831
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017763387297633872,
+      "loss": 0.1685,
+      "step": 11832
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0001776214196762142,
+      "loss": 0.2391,
+      "step": 11833
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017760896637608964,
+      "loss": 0.1459,
+      "step": 11834
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017759651307596512,
+      "loss": 0.2798,
+      "step": 11835
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017758405977584058,
+      "loss": 0.3303,
+      "step": 11836
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017757160647571604,
+      "loss": 0.2463,
+      "step": 11837
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017755915317559153,
+      "loss": 0.1633,
+      "step": 11838
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017754669987546696,
+      "loss": 0.1679,
+      "step": 11839
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017753424657534244,
+      "loss": 0.2392,
+      "step": 11840
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0001775217932752179,
+      "loss": 0.2473,
+      "step": 11841
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0001775093399750934,
+      "loss": 0.1762,
+      "step": 11842
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017749688667496885,
+      "loss": 0.3172,
+      "step": 11843
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017748443337484434,
+      "loss": 0.2759,
+      "step": 11844
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017747198007471977,
+      "loss": 0.1615,
+      "step": 11845
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017745952677459525,
+      "loss": 0.2267,
+      "step": 11846
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0001774470734744707,
+      "loss": 0.1707,
+      "step": 11847
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0001774346201743462,
+      "loss": 0.2149,
+      "step": 11848
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017742216687422166,
+      "loss": 0.4372,
+      "step": 11849
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017740971357409714,
+      "loss": 0.3059,
+      "step": 11850
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017739726027397258,
+      "loss": 0.2168,
+      "step": 11851
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017738480697384806,
+      "loss": 0.2648,
+      "step": 11852
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017737235367372352,
+      "loss": 0.2043,
+      "step": 11853
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017735990037359898,
+      "loss": 0.2427,
+      "step": 11854
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017734744707347447,
+      "loss": 0.2282,
+      "step": 11855
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.0001773349937733499,
+      "loss": 0.215,
+      "step": 11856
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00017732254047322539,
+      "loss": 0.1768,
+      "step": 11857
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017731008717310085,
+      "loss": 0.2518,
+      "step": 11858
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017729763387297633,
+      "loss": 0.2671,
+      "step": 11859
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0001772851805728518,
+      "loss": 0.2813,
+      "step": 11860
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017727272727272728,
+      "loss": 0.2528,
+      "step": 11861
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0001772602739726027,
+      "loss": 0.2313,
+      "step": 11862
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0001772478206724782,
+      "loss": 0.2029,
+      "step": 11863
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017723536737235365,
+      "loss": 0.1537,
+      "step": 11864
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017722291407222914,
+      "loss": 0.1799,
+      "step": 11865
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0001772104607721046,
+      "loss": 0.2239,
+      "step": 11866
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017719800747198009,
+      "loss": 0.3528,
+      "step": 11867
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017718555417185552,
+      "loss": 0.3019,
+      "step": 11868
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017717310087173098,
+      "loss": 0.2663,
+      "step": 11869
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017716064757160646,
+      "loss": 0.2041,
+      "step": 11870
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017714819427148192,
+      "loss": 0.1991,
+      "step": 11871
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0001771357409713574,
+      "loss": 0.3703,
+      "step": 11872
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017712328767123284,
+      "loss": 0.2061,
+      "step": 11873
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017711083437110833,
+      "loss": 0.3727,
+      "step": 11874
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0001770983810709838,
+      "loss": 0.2663,
+      "step": 11875
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017708592777085927,
+      "loss": 0.2828,
+      "step": 11876
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017707347447073473,
+      "loss": 0.2155,
+      "step": 11877
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017706102117061022,
+      "loss": 0.2128,
+      "step": 11878
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017704856787048565,
+      "loss": 0.22,
+      "step": 11879
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017703611457036114,
+      "loss": 0.3093,
+      "step": 11880
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0001770236612702366,
+      "loss": 0.2737,
+      "step": 11881
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017701120797011208,
+      "loss": 0.2579,
+      "step": 11882
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00017699875466998754,
+      "loss": 0.1818,
+      "step": 11883
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017698630136986297,
+      "loss": 0.1368,
+      "step": 11884
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017697384806973846,
+      "loss": 0.1079,
+      "step": 11885
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017696139476961392,
+      "loss": 0.3674,
+      "step": 11886
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0001769489414694894,
+      "loss": 0.2534,
+      "step": 11887
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017693648816936486,
+      "loss": 0.2072,
+      "step": 11888
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017692403486924035,
+      "loss": 0.1721,
+      "step": 11889
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017691158156911578,
+      "loss": 0.3969,
+      "step": 11890
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017689912826899127,
+      "loss": 0.1895,
+      "step": 11891
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017688667496886673,
+      "loss": 0.2657,
+      "step": 11892
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017687422166874221,
+      "loss": 0.3207,
+      "step": 11893
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017686176836861767,
+      "loss": 0.2517,
+      "step": 11894
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017684931506849316,
+      "loss": 0.2403,
+      "step": 11895
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0001768368617683686,
+      "loss": 0.189,
+      "step": 11896
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017682440846824408,
+      "loss": 0.2066,
+      "step": 11897
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017681195516811954,
+      "loss": 0.2177,
+      "step": 11898
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017679950186799502,
+      "loss": 0.4677,
+      "step": 11899
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017678704856787048,
+      "loss": 0.2084,
+      "step": 11900
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017677459526774591,
+      "loss": 0.1926,
+      "step": 11901
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0001767621419676214,
+      "loss": 0.1787,
+      "step": 11902
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017674968866749686,
+      "loss": 0.2193,
+      "step": 11903
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017673723536737235,
+      "loss": 0.1729,
+      "step": 11904
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0001767247820672478,
+      "loss": 0.2129,
+      "step": 11905
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0001767123287671233,
+      "loss": 0.1818,
+      "step": 11906
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017669987546699872,
+      "loss": 0.1967,
+      "step": 11907
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0001766874221668742,
+      "loss": 0.1216,
+      "step": 11908
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017667496886674967,
+      "loss": 0.1612,
+      "step": 11909
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00017666251556662516,
+      "loss": 0.1533,
+      "step": 11910
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017665006226650061,
+      "loss": 0.2095,
+      "step": 11911
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017663760896637607,
+      "loss": 0.1952,
+      "step": 11912
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017662515566625153,
+      "loss": 0.1647,
+      "step": 11913
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017661270236612702,
+      "loss": 0.1784,
+      "step": 11914
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017660024906600248,
+      "loss": 0.1743,
+      "step": 11915
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017658779576587794,
+      "loss": 0.1833,
+      "step": 11916
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017657534246575342,
+      "loss": 0.2103,
+      "step": 11917
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017656288916562886,
+      "loss": 0.3243,
+      "step": 11918
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017655043586550434,
+      "loss": 0.253,
+      "step": 11919
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0001765379825653798,
+      "loss": 0.171,
+      "step": 11920
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0001765255292652553,
+      "loss": 0.1994,
+      "step": 11921
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017651307596513075,
+      "loss": 0.1263,
+      "step": 11922
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0001765006226650062,
+      "loss": 0.3263,
+      "step": 11923
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017648816936488167,
+      "loss": 0.5699,
+      "step": 11924
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017647571606475715,
+      "loss": 0.1786,
+      "step": 11925
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0001764632627646326,
+      "loss": 0.2467,
+      "step": 11926
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0001764508094645081,
+      "loss": 0.1758,
+      "step": 11927
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017643835616438356,
+      "loss": 0.1765,
+      "step": 11928
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017642590286425902,
+      "loss": 0.3292,
+      "step": 11929
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017641344956413447,
+      "loss": 0.3091,
+      "step": 11930
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017640099626400993,
+      "loss": 0.1779,
+      "step": 11931
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017638854296388542,
+      "loss": 0.2698,
+      "step": 11932
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017637608966376088,
+      "loss": 0.2232,
+      "step": 11933
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017636363636363634,
+      "loss": 0.1603,
+      "step": 11934
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0001763511830635118,
+      "loss": 0.1618,
+      "step": 11935
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00017633872976338728,
+      "loss": 0.4176,
+      "step": 11936
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017632627646326274,
+      "loss": 0.3769,
+      "step": 11937
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017631382316313823,
+      "loss": 0.2297,
+      "step": 11938
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0001763013698630137,
+      "loss": 0.1311,
+      "step": 11939
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017628891656288915,
+      "loss": 0.167,
+      "step": 11940
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0001762764632627646,
+      "loss": 0.1251,
+      "step": 11941
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0001762640099626401,
+      "loss": 0.2514,
+      "step": 11942
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017625155666251555,
+      "loss": 0.3526,
+      "step": 11943
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017623910336239104,
+      "loss": 0.2384,
+      "step": 11944
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017622665006226647,
+      "loss": 0.2931,
+      "step": 11945
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017621419676214196,
+      "loss": 0.203,
+      "step": 11946
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017620174346201742,
+      "loss": 0.1691,
+      "step": 11947
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017618929016189288,
+      "loss": 0.2712,
+      "step": 11948
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017617683686176836,
+      "loss": 0.2455,
+      "step": 11949
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017616438356164382,
+      "loss": 0.2135,
+      "step": 11950
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017615193026151928,
+      "loss": 0.1605,
+      "step": 11951
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017613947696139474,
+      "loss": 0.166,
+      "step": 11952
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017612702366127023,
+      "loss": 0.282,
+      "step": 11953
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017611457036114568,
+      "loss": 0.2155,
+      "step": 11954
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017610211706102117,
+      "loss": 0.3431,
+      "step": 11955
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0001760896637608966,
+      "loss": 0.2638,
+      "step": 11956
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0001760772104607721,
+      "loss": 0.1723,
+      "step": 11957
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017606475716064755,
+      "loss": 0.1573,
+      "step": 11958
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017605230386052303,
+      "loss": 0.146,
+      "step": 11959
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0001760398505603985,
+      "loss": 0.2015,
+      "step": 11960
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00017602739726027398,
+      "loss": 0.2964,
+      "step": 11961
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.0001760149439601494,
+      "loss": 0.2211,
+      "step": 11962
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017600249066002487,
+      "loss": 0.2043,
+      "step": 11963
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017599003735990036,
+      "loss": 0.1431,
+      "step": 11964
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017597758405977582,
+      "loss": 0.1719,
+      "step": 11965
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0001759651307596513,
+      "loss": 0.2591,
+      "step": 11966
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017595267745952673,
+      "loss": 0.1751,
+      "step": 11967
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017594022415940222,
+      "loss": 0.2406,
+      "step": 11968
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017592777085927768,
+      "loss": 0.1535,
+      "step": 11969
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017591531755915317,
+      "loss": 0.1623,
+      "step": 11970
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017590286425902863,
+      "loss": 0.1908,
+      "step": 11971
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0001758904109589041,
+      "loss": 0.1416,
+      "step": 11972
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017587795765877954,
+      "loss": 0.2637,
+      "step": 11973
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017586550435865503,
+      "loss": 0.3357,
+      "step": 11974
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0001758530510585305,
+      "loss": 0.2107,
+      "step": 11975
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017584059775840598,
+      "loss": 0.2638,
+      "step": 11976
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017582814445828143,
+      "loss": 0.1331,
+      "step": 11977
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017581569115815692,
+      "loss": 0.2073,
+      "step": 11978
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017580323785803235,
+      "loss": 0.253,
+      "step": 11979
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0001757907845579078,
+      "loss": 0.2452,
+      "step": 11980
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0001757783312577833,
+      "loss": 0.2257,
+      "step": 11981
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017576587795765876,
+      "loss": 0.1927,
+      "step": 11982
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017575342465753424,
+      "loss": 0.2096,
+      "step": 11983
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017574097135740968,
+      "loss": 0.2405,
+      "step": 11984
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017572851805728516,
+      "loss": 0.1778,
+      "step": 11985
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017571606475716062,
+      "loss": 0.2852,
+      "step": 11986
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0001757036114570361,
+      "loss": 0.2439,
+      "step": 11987
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00017569115815691157,
+      "loss": 0.2,
+      "step": 11988
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017567870485678705,
+      "loss": 0.2727,
+      "step": 11989
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017566625155666249,
+      "loss": 0.2037,
+      "step": 11990
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017565379825653797,
+      "loss": 0.2044,
+      "step": 11991
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017564134495641343,
+      "loss": 0.3413,
+      "step": 11992
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017562889165628892,
+      "loss": 0.4049,
+      "step": 11993
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017561643835616438,
+      "loss": 0.2419,
+      "step": 11994
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0001756039850560398,
+      "loss": 0.1961,
+      "step": 11995
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0001755915317559153,
+      "loss": 0.1815,
+      "step": 11996
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017557907845579075,
+      "loss": 0.1612,
+      "step": 11997
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017556662515566624,
+      "loss": 0.2821,
+      "step": 11998
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0001755541718555417,
+      "loss": 0.3581,
+      "step": 11999
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017554171855541719,
+      "loss": 0.25,
+      "step": 12000
+    },
+    {
+      "epoch": 4.6,
+      "eval_cer": 0.10522660319566109,
+      "eval_loss": 0.37344011664390564,
+      "eval_runtime": 47.2334,
+      "eval_samples_per_second": 37.114,
+      "eval_steps_per_second": 4.658,
+      "eval_wer": 0.17228583065574513,
+      "step": 12000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017552926525529262,
+      "loss": 0.2239,
+      "step": 12001
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0001755168119551681,
+      "loss": 0.1577,
+      "step": 12002
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017550435865504356,
+      "loss": 0.192,
+      "step": 12003
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017549190535491905,
+      "loss": 0.1987,
+      "step": 12004
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0001754794520547945,
+      "loss": 0.4291,
+      "step": 12005
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017546699875467,
+      "loss": 0.1928,
+      "step": 12006
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017545454545454543,
+      "loss": 0.2836,
+      "step": 12007
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0001754420921544209,
+      "loss": 0.1911,
+      "step": 12008
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017542963885429637,
+      "loss": 0.1823,
+      "step": 12009
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017541718555417183,
+      "loss": 0.3151,
+      "step": 12010
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017540473225404732,
+      "loss": 0.2918,
+      "step": 12011
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017539227895392275,
+      "loss": 0.2754,
+      "step": 12012
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00017537982565379824,
+      "loss": 0.2152,
+      "step": 12013
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0001753673723536737,
+      "loss": 0.1783,
+      "step": 12014
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017535491905354918,
+      "loss": 0.2016,
+      "step": 12015
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017534246575342464,
+      "loss": 0.2039,
+      "step": 12016
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017533001245330013,
+      "loss": 0.1949,
+      "step": 12017
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017531755915317556,
+      "loss": 0.4439,
+      "step": 12018
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017530510585305105,
+      "loss": 0.2728,
+      "step": 12019
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001752926525529265,
+      "loss": 0.2796,
+      "step": 12020
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.000175280199252802,
+      "loss": 0.2563,
+      "step": 12021
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017526774595267745,
+      "loss": 0.1653,
+      "step": 12022
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017525529265255294,
+      "loss": 0.2824,
+      "step": 12023
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017524283935242837,
+      "loss": 0.3648,
+      "step": 12024
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017523038605230385,
+      "loss": 0.3013,
+      "step": 12025
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017521793275217931,
+      "loss": 0.2047,
+      "step": 12026
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017520547945205477,
+      "loss": 0.1311,
+      "step": 12027
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017519302615193026,
+      "loss": 0.2047,
+      "step": 12028
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001751805728518057,
+      "loss": 0.235,
+      "step": 12029
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017516811955168118,
+      "loss": 0.2617,
+      "step": 12030
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017515566625155664,
+      "loss": 0.1791,
+      "step": 12031
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017514321295143212,
+      "loss": 0.1936,
+      "step": 12032
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017513075965130758,
+      "loss": 0.1738,
+      "step": 12033
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017511830635118307,
+      "loss": 0.1497,
+      "step": 12034
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001751058530510585,
+      "loss": 0.1812,
+      "step": 12035
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.000175093399750934,
+      "loss": 0.3634,
+      "step": 12036
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017508094645080945,
+      "loss": 0.2404,
+      "step": 12037
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017506849315068493,
+      "loss": 0.2705,
+      "step": 12038
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001750560398505604,
+      "loss": 0.2269,
+      "step": 12039
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00017504358655043588,
+      "loss": 0.1861,
+      "step": 12040
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0001750311332503113,
+      "loss": 0.179,
+      "step": 12041
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017501867995018677,
+      "loss": 0.252,
+      "step": 12042
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017500622665006226,
+      "loss": 0.2661,
+      "step": 12043
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017499377334993771,
+      "loss": 0.1961,
+      "step": 12044
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0001749813200498132,
+      "loss": 0.2841,
+      "step": 12045
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017496886674968863,
+      "loss": 0.3629,
+      "step": 12046
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017495641344956412,
+      "loss": 0.2114,
+      "step": 12047
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017494396014943958,
+      "loss": 0.1308,
+      "step": 12048
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017493150684931506,
+      "loss": 0.2575,
+      "step": 12049
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017491905354919052,
+      "loss": 0.1838,
+      "step": 12050
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.000174906600249066,
+      "loss": 0.1754,
+      "step": 12051
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017489414694894144,
+      "loss": 0.1417,
+      "step": 12052
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017488169364881693,
+      "loss": 0.1741,
+      "step": 12053
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0001748692403486924,
+      "loss": 0.2192,
+      "step": 12054
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017485678704856787,
+      "loss": 0.277,
+      "step": 12055
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017484433374844333,
+      "loss": 0.1905,
+      "step": 12056
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017483188044831876,
+      "loss": 0.2257,
+      "step": 12057
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017481942714819425,
+      "loss": 0.1776,
+      "step": 12058
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0001748069738480697,
+      "loss": 0.2632,
+      "step": 12059
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0001747945205479452,
+      "loss": 0.3488,
+      "step": 12060
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017478206724782066,
+      "loss": 0.2702,
+      "step": 12061
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017476961394769614,
+      "loss": 0.2527,
+      "step": 12062
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017475716064757157,
+      "loss": 0.3851,
+      "step": 12063
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017474470734744706,
+      "loss": 0.2558,
+      "step": 12064
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017473225404732252,
+      "loss": 0.1958,
+      "step": 12065
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.000174719800747198,
+      "loss": 0.127,
+      "step": 12066
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017470734744707346,
+      "loss": 0.1738,
+      "step": 12067
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017469489414694895,
+      "loss": 0.356,
+      "step": 12068
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017468244084682438,
+      "loss": 0.206,
+      "step": 12069
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017466998754669987,
+      "loss": 0.3118,
+      "step": 12070
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017465753424657533,
+      "loss": 0.1973,
+      "step": 12071
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017464508094645081,
+      "loss": 0.1688,
+      "step": 12072
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017463262764632627,
+      "loss": 0.1568,
+      "step": 12073
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0001746201743462017,
+      "loss": 0.2823,
+      "step": 12074
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0001746077210460772,
+      "loss": 0.1842,
+      "step": 12075
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017459526774595265,
+      "loss": 0.2223,
+      "step": 12076
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017458281444582814,
+      "loss": 0.2084,
+      "step": 12077
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0001745703611457036,
+      "loss": 0.1837,
+      "step": 12078
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017455790784557908,
+      "loss": 0.1693,
+      "step": 12079
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017454545454545452,
+      "loss": 0.2568,
+      "step": 12080
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017453300124533,
+      "loss": 0.2396,
+      "step": 12081
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017452054794520546,
+      "loss": 0.2264,
+      "step": 12082
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017450809464508095,
+      "loss": 0.1337,
+      "step": 12083
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0001744956413449564,
+      "loss": 0.1596,
+      "step": 12084
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017448318804483187,
+      "loss": 0.1891,
+      "step": 12085
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017447073474470732,
+      "loss": 0.3802,
+      "step": 12086
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0001744582814445828,
+      "loss": 0.2243,
+      "step": 12087
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017444582814445827,
+      "loss": 0.2383,
+      "step": 12088
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017443337484433373,
+      "loss": 0.1996,
+      "step": 12089
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017442092154420922,
+      "loss": 0.1466,
+      "step": 12090
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017440846824408465,
+      "loss": 0.2213,
+      "step": 12091
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00017439601494396013,
+      "loss": 0.1935,
+      "step": 12092
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0001743835616438356,
+      "loss": 0.2345,
+      "step": 12093
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017437110834371108,
+      "loss": 0.2343,
+      "step": 12094
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017435865504358654,
+      "loss": 0.2035,
+      "step": 12095
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.000174346201743462,
+      "loss": 0.174,
+      "step": 12096
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017433374844333746,
+      "loss": 0.2632,
+      "step": 12097
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017432129514321294,
+      "loss": 0.1846,
+      "step": 12098
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0001743088418430884,
+      "loss": 0.3233,
+      "step": 12099
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0001742963885429639,
+      "loss": 0.3544,
+      "step": 12100
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017428393524283935,
+      "loss": 0.1894,
+      "step": 12101
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0001742714819427148,
+      "loss": 0.1711,
+      "step": 12102
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017425902864259027,
+      "loss": 0.2175,
+      "step": 12103
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017424657534246573,
+      "loss": 0.2383,
+      "step": 12104
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0001742341220423412,
+      "loss": 0.3341,
+      "step": 12105
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017422166874221667,
+      "loss": 0.2188,
+      "step": 12106
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017420921544209213,
+      "loss": 0.2136,
+      "step": 12107
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0001741967621419676,
+      "loss": 0.318,
+      "step": 12108
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017418430884184308,
+      "loss": 0.1412,
+      "step": 12109
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017417185554171853,
+      "loss": 0.2974,
+      "step": 12110
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017415940224159402,
+      "loss": 0.3215,
+      "step": 12111
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017414694894146948,
+      "loss": 0.2422,
+      "step": 12112
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017413449564134494,
+      "loss": 0.1934,
+      "step": 12113
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0001741220423412204,
+      "loss": 0.2349,
+      "step": 12114
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017410958904109588,
+      "loss": 0.1841,
+      "step": 12115
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017409713574097134,
+      "loss": 0.2071,
+      "step": 12116
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017408468244084683,
+      "loss": 0.2339,
+      "step": 12117
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00017407222914072226,
+      "loss": 0.3743,
+      "step": 12118
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017405977584059775,
+      "loss": 0.2006,
+      "step": 12119
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0001740473225404732,
+      "loss": 0.2069,
+      "step": 12120
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017403486924034867,
+      "loss": 0.1579,
+      "step": 12121
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017402241594022415,
+      "loss": 0.2351,
+      "step": 12122
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017400996264009958,
+      "loss": 0.2848,
+      "step": 12123
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017399750933997507,
+      "loss": 0.3456,
+      "step": 12124
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017398505603985053,
+      "loss": 0.359,
+      "step": 12125
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017397260273972602,
+      "loss": 0.2566,
+      "step": 12126
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017396014943960148,
+      "loss": 0.1541,
+      "step": 12127
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017394769613947696,
+      "loss": 0.2854,
+      "step": 12128
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0001739352428393524,
+      "loss": 0.2138,
+      "step": 12129
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017392278953922788,
+      "loss": 0.254,
+      "step": 12130
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017391033623910334,
+      "loss": 0.2279,
+      "step": 12131
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017389788293897883,
+      "loss": 0.2398,
+      "step": 12132
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017388542963885429,
+      "loss": 0.1758,
+      "step": 12133
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017387297633872977,
+      "loss": 0.1552,
+      "step": 12134
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0001738605230386052,
+      "loss": 0.3537,
+      "step": 12135
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017384806973848066,
+      "loss": 0.3984,
+      "step": 12136
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017383561643835615,
+      "loss": 0.1806,
+      "step": 12137
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0001738231631382316,
+      "loss": 0.1916,
+      "step": 12138
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0001738107098381071,
+      "loss": 0.2629,
+      "step": 12139
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017379825653798253,
+      "loss": 0.1749,
+      "step": 12140
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.000173785803237858,
+      "loss": 0.178,
+      "step": 12141
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017377334993773347,
+      "loss": 0.2473,
+      "step": 12142
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017376089663760896,
+      "loss": 0.2445,
+      "step": 12143
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00017374844333748442,
+      "loss": 0.2494,
+      "step": 12144
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0001737359900373599,
+      "loss": 0.2008,
+      "step": 12145
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017372353673723534,
+      "loss": 0.2127,
+      "step": 12146
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017371108343711082,
+      "loss": 0.2206,
+      "step": 12147
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017369863013698628,
+      "loss": 0.1639,
+      "step": 12148
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017368617683686177,
+      "loss": 0.2195,
+      "step": 12149
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017367372353673723,
+      "loss": 0.2029,
+      "step": 12150
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017366127023661266,
+      "loss": 0.1967,
+      "step": 12151
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017364881693648814,
+      "loss": 0.1845,
+      "step": 12152
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0001736363636363636,
+      "loss": 0.1666,
+      "step": 12153
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0001736239103362391,
+      "loss": 0.2916,
+      "step": 12154
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017361145703611455,
+      "loss": 0.2233,
+      "step": 12155
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017359900373599004,
+      "loss": 0.4798,
+      "step": 12156
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017358655043586547,
+      "loss": 0.2966,
+      "step": 12157
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017357409713574095,
+      "loss": 0.2347,
+      "step": 12158
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0001735616438356164,
+      "loss": 0.3418,
+      "step": 12159
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0001735491905354919,
+      "loss": 0.1885,
+      "step": 12160
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017353673723536736,
+      "loss": 0.3388,
+      "step": 12161
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017352428393524284,
+      "loss": 0.2249,
+      "step": 12162
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017351183063511828,
+      "loss": 0.1712,
+      "step": 12163
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017349937733499376,
+      "loss": 0.1768,
+      "step": 12164
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017348692403486922,
+      "loss": 0.1318,
+      "step": 12165
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0001734744707347447,
+      "loss": 0.1464,
+      "step": 12166
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017346201743462017,
+      "loss": 0.2786,
+      "step": 12167
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0001734495641344956,
+      "loss": 0.2317,
+      "step": 12168
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017343711083437109,
+      "loss": 0.1946,
+      "step": 12169
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00017342465753424655,
+      "loss": 0.1628,
+      "step": 12170
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017341220423412203,
+      "loss": 0.3176,
+      "step": 12171
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0001733997509339975,
+      "loss": 0.1998,
+      "step": 12172
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017338729763387298,
+      "loss": 0.2733,
+      "step": 12173
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0001733748443337484,
+      "loss": 0.4062,
+      "step": 12174
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0001733623910336239,
+      "loss": 0.1753,
+      "step": 12175
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017334993773349935,
+      "loss": 0.3041,
+      "step": 12176
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017333748443337484,
+      "loss": 0.2257,
+      "step": 12177
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0001733250311332503,
+      "loss": 0.2655,
+      "step": 12178
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017331257783312579,
+      "loss": 0.2647,
+      "step": 12179
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017330012453300122,
+      "loss": 0.3799,
+      "step": 12180
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0001732876712328767,
+      "loss": 0.2283,
+      "step": 12181
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017327521793275216,
+      "loss": 0.2452,
+      "step": 12182
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017326276463262762,
+      "loss": 0.1761,
+      "step": 12183
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0001732503113325031,
+      "loss": 0.1524,
+      "step": 12184
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017323785803237854,
+      "loss": 0.2246,
+      "step": 12185
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017322540473225403,
+      "loss": 0.3766,
+      "step": 12186
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0001732129514321295,
+      "loss": 0.2003,
+      "step": 12187
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017320049813200497,
+      "loss": 0.1638,
+      "step": 12188
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017318804483188043,
+      "loss": 0.3172,
+      "step": 12189
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017317559153175592,
+      "loss": 0.1852,
+      "step": 12190
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017316313823163135,
+      "loss": 0.1871,
+      "step": 12191
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017315068493150684,
+      "loss": 0.2002,
+      "step": 12192
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0001731382316313823,
+      "loss": 0.3146,
+      "step": 12193
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017312577833125778,
+      "loss": 0.1519,
+      "step": 12194
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017311332503113324,
+      "loss": 0.1752,
+      "step": 12195
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017310087173100873,
+      "loss": 0.1885,
+      "step": 12196
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00017308841843088416,
+      "loss": 0.1266,
+      "step": 12197
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017307596513075962,
+      "loss": 0.211,
+      "step": 12198
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0001730635118306351,
+      "loss": 0.3827,
+      "step": 12199
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017305105853051056,
+      "loss": 0.2049,
+      "step": 12200
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017303860523038605,
+      "loss": 0.1443,
+      "step": 12201
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017302615193026148,
+      "loss": 0.2334,
+      "step": 12202
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017301369863013697,
+      "loss": 0.1507,
+      "step": 12203
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017300124533001243,
+      "loss": 0.2019,
+      "step": 12204
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017298879202988791,
+      "loss": 0.3563,
+      "step": 12205
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017297633872976337,
+      "loss": 0.1698,
+      "step": 12206
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017296388542963886,
+      "loss": 0.2833,
+      "step": 12207
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0001729514321295143,
+      "loss": 0.1655,
+      "step": 12208
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017293897882938978,
+      "loss": 0.1882,
+      "step": 12209
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017292652552926524,
+      "loss": 0.2934,
+      "step": 12210
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017291407222914072,
+      "loss": 0.353,
+      "step": 12211
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017290161892901618,
+      "loss": 0.2012,
+      "step": 12212
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017288916562889167,
+      "loss": 0.228,
+      "step": 12213
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0001728767123287671,
+      "loss": 0.1183,
+      "step": 12214
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017286425902864256,
+      "loss": 0.1851,
+      "step": 12215
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017285180572851805,
+      "loss": 0.1612,
+      "step": 12216
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0001728393524283935,
+      "loss": 0.1978,
+      "step": 12217
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.000172826899128269,
+      "loss": 0.4064,
+      "step": 12218
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017281444582814442,
+      "loss": 0.1867,
+      "step": 12219
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.0001728019925280199,
+      "loss": 0.2354,
+      "step": 12220
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017278953922789537,
+      "loss": 0.144,
+      "step": 12221
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017277708592777086,
+      "loss": 0.1917,
+      "step": 12222
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00017276463262764631,
+      "loss": 0.2726,
+      "step": 12223
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0001727521793275218,
+      "loss": 0.3287,
+      "step": 12224
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017273972602739723,
+      "loss": 0.2528,
+      "step": 12225
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017272727272727272,
+      "loss": 0.1836,
+      "step": 12226
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017271481942714818,
+      "loss": 0.2891,
+      "step": 12227
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017270236612702366,
+      "loss": 0.231,
+      "step": 12228
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017268991282689912,
+      "loss": 0.1557,
+      "step": 12229
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017267745952677456,
+      "loss": 0.1848,
+      "step": 12230
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017266500622665004,
+      "loss": 0.1904,
+      "step": 12231
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0001726525529265255,
+      "loss": 0.2185,
+      "step": 12232
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.000172640099626401,
+      "loss": 0.1697,
+      "step": 12233
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017262764632627645,
+      "loss": 0.2255,
+      "step": 12234
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017261519302615193,
+      "loss": 0.1681,
+      "step": 12235
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017260273972602737,
+      "loss": 0.2627,
+      "step": 12236
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017259028642590285,
+      "loss": 0.2562,
+      "step": 12237
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0001725778331257783,
+      "loss": 0.1627,
+      "step": 12238
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0001725653798256538,
+      "loss": 0.1658,
+      "step": 12239
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017255292652552926,
+      "loss": 0.1862,
+      "step": 12240
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017254047322540474,
+      "loss": 0.2524,
+      "step": 12241
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017252801992528017,
+      "loss": 0.2945,
+      "step": 12242
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017251556662515566,
+      "loss": 0.3251,
+      "step": 12243
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017250311332503112,
+      "loss": 0.4096,
+      "step": 12244
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0001724906600249066,
+      "loss": 0.1489,
+      "step": 12245
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017247820672478207,
+      "loss": 0.2315,
+      "step": 12246
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0001724657534246575,
+      "loss": 0.2223,
+      "step": 12247
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017245330012453298,
+      "loss": 0.2468,
+      "step": 12248
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00017244084682440844,
+      "loss": 0.2207,
+      "step": 12249
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017242839352428393,
+      "loss": 0.1931,
+      "step": 12250
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0001724159402241594,
+      "loss": 0.256,
+      "step": 12251
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017240348692403487,
+      "loss": 0.2608,
+      "step": 12252
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0001723910336239103,
+      "loss": 0.164,
+      "step": 12253
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0001723785803237858,
+      "loss": 0.1464,
+      "step": 12254
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017236612702366125,
+      "loss": 0.3679,
+      "step": 12255
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017235367372353674,
+      "loss": 0.2873,
+      "step": 12256
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0001723412204234122,
+      "loss": 0.2015,
+      "step": 12257
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017232876712328766,
+      "loss": 0.185,
+      "step": 12258
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017231631382316312,
+      "loss": 0.2856,
+      "step": 12259
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0001723038605230386,
+      "loss": 0.2517,
+      "step": 12260
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017229140722291406,
+      "loss": 0.3083,
+      "step": 12261
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017227895392278952,
+      "loss": 0.2168,
+      "step": 12262
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.000172266500622665,
+      "loss": 0.4157,
+      "step": 12263
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017225404732254044,
+      "loss": 0.2954,
+      "step": 12264
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017224159402241593,
+      "loss": 0.2168,
+      "step": 12265
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017222914072229138,
+      "loss": 0.3253,
+      "step": 12266
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017221668742216687,
+      "loss": 0.2793,
+      "step": 12267
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017220423412204233,
+      "loss": 0.3796,
+      "step": 12268
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0001721917808219178,
+      "loss": 0.1922,
+      "step": 12269
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017217932752179325,
+      "loss": 0.212,
+      "step": 12270
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017216687422166873,
+      "loss": 0.2004,
+      "step": 12271
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0001721544209215442,
+      "loss": 0.2666,
+      "step": 12272
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00017214196762141968,
+      "loss": 0.1731,
+      "step": 12273
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0001721295143212951,
+      "loss": 0.3533,
+      "step": 12274
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0001721170610211706,
+      "loss": 0.2497,
+      "step": 12275
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017210460772104606,
+      "loss": 0.1502,
+      "step": 12276
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017209215442092152,
+      "loss": 0.1315,
+      "step": 12277
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.000172079701120797,
+      "loss": 0.2201,
+      "step": 12278
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017206724782067246,
+      "loss": 0.2448,
+      "step": 12279
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017205479452054792,
+      "loss": 0.3625,
+      "step": 12280
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017204234122042338,
+      "loss": 0.2676,
+      "step": 12281
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017202988792029887,
+      "loss": 0.175,
+      "step": 12282
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017201743462017433,
+      "loss": 0.2094,
+      "step": 12283
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0001720049813200498,
+      "loss": 0.2667,
+      "step": 12284
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017199252801992524,
+      "loss": 0.1481,
+      "step": 12285
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017198007471980073,
+      "loss": 0.3115,
+      "step": 12286
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0001719676214196762,
+      "loss": 0.2273,
+      "step": 12287
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017195516811955168,
+      "loss": 0.3273,
+      "step": 12288
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017194271481942714,
+      "loss": 0.2558,
+      "step": 12289
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017193026151930262,
+      "loss": 0.2244,
+      "step": 12290
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017191780821917805,
+      "loss": 0.1817,
+      "step": 12291
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017190535491905354,
+      "loss": 0.1807,
+      "step": 12292
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.000171892901618929,
+      "loss": 0.2534,
+      "step": 12293
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017188044831880446,
+      "loss": 0.2381,
+      "step": 12294
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017186799501867994,
+      "loss": 0.2219,
+      "step": 12295
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017185554171855538,
+      "loss": 0.1891,
+      "step": 12296
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017184308841843086,
+      "loss": 0.0921,
+      "step": 12297
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017183063511830632,
+      "loss": 0.1662,
+      "step": 12298
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0001718181818181818,
+      "loss": 0.3658,
+      "step": 12299
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017180572851805727,
+      "loss": 0.2621,
+      "step": 12300
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00017179327521793275,
+      "loss": 0.2596,
+      "step": 12301
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017178082191780819,
+      "loss": 0.2596,
+      "step": 12302
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017176836861768367,
+      "loss": 0.4232,
+      "step": 12303
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017175591531755913,
+      "loss": 0.1968,
+      "step": 12304
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017174346201743462,
+      "loss": 0.3594,
+      "step": 12305
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017173100871731008,
+      "loss": 0.1768,
+      "step": 12306
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017171855541718556,
+      "loss": 0.199,
+      "step": 12307
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.000171706102117061,
+      "loss": 0.1607,
+      "step": 12308
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017169364881693645,
+      "loss": 0.2491,
+      "step": 12309
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017168119551681194,
+      "loss": 0.176,
+      "step": 12310
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0001716687422166874,
+      "loss": 0.387,
+      "step": 12311
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017165628891656289,
+      "loss": 0.2166,
+      "step": 12312
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017164383561643832,
+      "loss": 0.3356,
+      "step": 12313
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0001716313823163138,
+      "loss": 0.1757,
+      "step": 12314
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017161892901618926,
+      "loss": 0.1984,
+      "step": 12315
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017160647571606475,
+      "loss": 0.1546,
+      "step": 12316
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0001715940224159402,
+      "loss": 0.2473,
+      "step": 12317
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0001715815691158157,
+      "loss": 0.2516,
+      "step": 12318
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017156911581569113,
+      "loss": 0.2017,
+      "step": 12319
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0001715566625155666,
+      "loss": 0.1791,
+      "step": 12320
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017154420921544207,
+      "loss": 0.1369,
+      "step": 12321
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017153175591531756,
+      "loss": 0.3333,
+      "step": 12322
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017151930261519302,
+      "loss": 0.1554,
+      "step": 12323
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017150684931506845,
+      "loss": 0.4211,
+      "step": 12324
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017149439601494394,
+      "loss": 0.3084,
+      "step": 12325
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0001714819427148194,
+      "loss": 0.2876,
+      "step": 12326
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00017146948941469488,
+      "loss": 0.2173,
+      "step": 12327
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017145703611457034,
+      "loss": 0.2197,
+      "step": 12328
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017144458281444583,
+      "loss": 0.2374,
+      "step": 12329
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017143212951432126,
+      "loss": 0.3124,
+      "step": 12330
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017141967621419675,
+      "loss": 0.3364,
+      "step": 12331
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0001714072229140722,
+      "loss": 0.1881,
+      "step": 12332
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0001713947696139477,
+      "loss": 0.266,
+      "step": 12333
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017138231631382315,
+      "loss": 0.2149,
+      "step": 12334
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017136986301369864,
+      "loss": 0.321,
+      "step": 12335
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017135740971357407,
+      "loss": 0.3783,
+      "step": 12336
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017134495641344955,
+      "loss": 0.2149,
+      "step": 12337
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017133250311332501,
+      "loss": 0.2747,
+      "step": 12338
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0001713200498132005,
+      "loss": 0.1955,
+      "step": 12339
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017130759651307596,
+      "loss": 0.2696,
+      "step": 12340
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0001712951432129514,
+      "loss": 0.1634,
+      "step": 12341
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017128268991282688,
+      "loss": 0.186,
+      "step": 12342
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017127023661270234,
+      "loss": 0.3901,
+      "step": 12343
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017125778331257782,
+      "loss": 0.1854,
+      "step": 12344
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017124533001245328,
+      "loss": 0.173,
+      "step": 12345
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017123287671232877,
+      "loss": 0.1884,
+      "step": 12346
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0001712204234122042,
+      "loss": 0.2558,
+      "step": 12347
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0001712079701120797,
+      "loss": 0.2132,
+      "step": 12348
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017119551681195515,
+      "loss": 0.3677,
+      "step": 12349
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017118306351183063,
+      "loss": 0.2336,
+      "step": 12350
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0001711706102117061,
+      "loss": 0.2301,
+      "step": 12351
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00017115815691158158,
+      "loss": 0.2114,
+      "step": 12352
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.000171145703611457,
+      "loss": 0.2343,
+      "step": 12353
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001711332503113325,
+      "loss": 0.1611,
+      "step": 12354
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017112079701120796,
+      "loss": 0.2685,
+      "step": 12355
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017110834371108341,
+      "loss": 0.3781,
+      "step": 12356
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001710958904109589,
+      "loss": 0.1604,
+      "step": 12357
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017108343711083433,
+      "loss": 0.2923,
+      "step": 12358
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017107098381070982,
+      "loss": 0.2186,
+      "step": 12359
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017105853051058528,
+      "loss": 0.2043,
+      "step": 12360
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017104607721046076,
+      "loss": 0.3361,
+      "step": 12361
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017103362391033622,
+      "loss": 0.1887,
+      "step": 12362
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001710211706102117,
+      "loss": 0.277,
+      "step": 12363
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017100871731008714,
+      "loss": 0.1764,
+      "step": 12364
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017099626400996263,
+      "loss": 0.1832,
+      "step": 12365
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001709838107098381,
+      "loss": 0.181,
+      "step": 12366
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017097135740971357,
+      "loss": 0.2318,
+      "step": 12367
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017095890410958903,
+      "loss": 0.2683,
+      "step": 12368
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017094645080946452,
+      "loss": 0.28,
+      "step": 12369
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017093399750933995,
+      "loss": 0.2127,
+      "step": 12370
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001709215442092154,
+      "loss": 0.1654,
+      "step": 12371
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001709090909090909,
+      "loss": 0.1657,
+      "step": 12372
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017089663760896636,
+      "loss": 0.162,
+      "step": 12373
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017088418430884184,
+      "loss": 0.3253,
+      "step": 12374
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017087173100871727,
+      "loss": 0.2363,
+      "step": 12375
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017085927770859276,
+      "loss": 0.3419,
+      "step": 12376
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017084682440846822,
+      "loss": 0.326,
+      "step": 12377
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001708343711083437,
+      "loss": 0.1594,
+      "step": 12378
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00017082191780821916,
+      "loss": 0.1982,
+      "step": 12379
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017080946450809465,
+      "loss": 0.2735,
+      "step": 12380
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017079701120797008,
+      "loss": 0.2622,
+      "step": 12381
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017078455790784557,
+      "loss": 0.2372,
+      "step": 12382
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017077210460772103,
+      "loss": 0.1453,
+      "step": 12383
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017075965130759652,
+      "loss": 0.1241,
+      "step": 12384
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017074719800747197,
+      "loss": 0.2486,
+      "step": 12385
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017073474470734746,
+      "loss": 0.4506,
+      "step": 12386
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001707222914072229,
+      "loss": 0.1994,
+      "step": 12387
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017070983810709835,
+      "loss": 0.1442,
+      "step": 12388
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017069738480697384,
+      "loss": 0.2482,
+      "step": 12389
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001706849315068493,
+      "loss": 0.1512,
+      "step": 12390
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017067247820672478,
+      "loss": 0.1262,
+      "step": 12391
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017066002490660022,
+      "loss": 0.2834,
+      "step": 12392
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001706475716064757,
+      "loss": 0.5093,
+      "step": 12393
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017063511830635116,
+      "loss": 0.1808,
+      "step": 12394
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017062266500622665,
+      "loss": 0.2065,
+      "step": 12395
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001706102117061021,
+      "loss": 0.3834,
+      "step": 12396
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001705977584059776,
+      "loss": 0.2187,
+      "step": 12397
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017058530510585302,
+      "loss": 0.2726,
+      "step": 12398
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001705728518057285,
+      "loss": 0.3384,
+      "step": 12399
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017056039850560397,
+      "loss": 0.2415,
+      "step": 12400
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017054794520547946,
+      "loss": 0.2765,
+      "step": 12401
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017053549190535492,
+      "loss": 0.1623,
+      "step": 12402
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017052303860523035,
+      "loss": 0.2834,
+      "step": 12403
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00017051058530510583,
+      "loss": 0.1328,
+      "step": 12404
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001704981320049813,
+      "loss": 0.3063,
+      "step": 12405
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017048567870485678,
+      "loss": 0.1656,
+      "step": 12406
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017047322540473224,
+      "loss": 0.1746,
+      "step": 12407
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017046077210460772,
+      "loss": 0.1683,
+      "step": 12408
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017044831880448316,
+      "loss": 0.3107,
+      "step": 12409
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017043586550435864,
+      "loss": 0.2855,
+      "step": 12410
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0001704234122042341,
+      "loss": 0.3628,
+      "step": 12411
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0001704109589041096,
+      "loss": 0.2081,
+      "step": 12412
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017039850560398505,
+      "loss": 0.1956,
+      "step": 12413
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017038605230386053,
+      "loss": 0.2346,
+      "step": 12414
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017037359900373597,
+      "loss": 0.1948,
+      "step": 12415
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017036114570361145,
+      "loss": 0.177,
+      "step": 12416
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0001703486924034869,
+      "loss": 0.426,
+      "step": 12417
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017033623910336237,
+      "loss": 0.3552,
+      "step": 12418
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017032378580323786,
+      "loss": 0.2382,
+      "step": 12419
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0001703113325031133,
+      "loss": 0.342,
+      "step": 12420
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017029887920298878,
+      "loss": 0.2634,
+      "step": 12421
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017028642590286423,
+      "loss": 0.2113,
+      "step": 12422
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017027397260273972,
+      "loss": 0.2205,
+      "step": 12423
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017026151930261518,
+      "loss": 0.3124,
+      "step": 12424
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017024906600249067,
+      "loss": 0.2504,
+      "step": 12425
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0001702366127023661,
+      "loss": 0.2277,
+      "step": 12426
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017022415940224158,
+      "loss": 0.2627,
+      "step": 12427
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017021170610211704,
+      "loss": 0.2727,
+      "step": 12428
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017019925280199253,
+      "loss": 0.181,
+      "step": 12429
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.000170186799501868,
+      "loss": 0.3142,
+      "step": 12430
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00017017434620174345,
+      "loss": 0.2268,
+      "step": 12431
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0001701618929016189,
+      "loss": 0.1902,
+      "step": 12432
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0001701494396014944,
+      "loss": 0.1845,
+      "step": 12433
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00017013698630136985,
+      "loss": 0.1306,
+      "step": 12434
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0001701245330012453,
+      "loss": 0.1979,
+      "step": 12435
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00017011207970112077,
+      "loss": 0.1781,
+      "step": 12436
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00017009962640099623,
+      "loss": 0.2148,
+      "step": 12437
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00017008717310087172,
+      "loss": 0.2778,
+      "step": 12438
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00017007471980074718,
+      "loss": 0.2246,
+      "step": 12439
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00017006226650062266,
+      "loss": 0.136,
+      "step": 12440
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00017004981320049812,
+      "loss": 0.3006,
+      "step": 12441
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00017003735990037358,
+      "loss": 0.2659,
+      "step": 12442
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00017002490660024904,
+      "loss": 0.242,
+      "step": 12443
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00017001245330012453,
+      "loss": 0.2664,
+      "step": 12444
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00016999999999999999,
+      "loss": 0.3735,
+      "step": 12445
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00016998754669987547,
+      "loss": 0.2433,
+      "step": 12446
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0001699750933997509,
+      "loss": 0.1718,
+      "step": 12447
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0001699626400996264,
+      "loss": 0.3292,
+      "step": 12448
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00016995018679950185,
+      "loss": 0.2816,
+      "step": 12449
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0001699377334993773,
+      "loss": 0.219,
+      "step": 12450
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0001699252801992528,
+      "loss": 0.3067,
+      "step": 12451
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00016991282689912825,
+      "loss": 0.2019,
+      "step": 12452
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0001699003735990037,
+      "loss": 0.2719,
+      "step": 12453
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00016988792029887917,
+      "loss": 0.208,
+      "step": 12454
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00016987546699875466,
+      "loss": 0.2984,
+      "step": 12455
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00016986301369863012,
+      "loss": 0.2133,
+      "step": 12456
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0001698505603985056,
+      "loss": 0.2426,
+      "step": 12457
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016983810709838104,
+      "loss": 0.3009,
+      "step": 12458
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016982565379825652,
+      "loss": 0.2071,
+      "step": 12459
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016981320049813198,
+      "loss": 0.1574,
+      "step": 12460
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016980074719800747,
+      "loss": 0.3392,
+      "step": 12461
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016978829389788293,
+      "loss": 0.181,
+      "step": 12462
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0001697758405977584,
+      "loss": 0.3266,
+      "step": 12463
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016976338729763384,
+      "loss": 0.1603,
+      "step": 12464
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0001697509339975093,
+      "loss": 0.1943,
+      "step": 12465
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0001697384806973848,
+      "loss": 0.1872,
+      "step": 12466
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016972602739726025,
+      "loss": 0.2806,
+      "step": 12467
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016971357409713574,
+      "loss": 0.3284,
+      "step": 12468
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016970112079701117,
+      "loss": 0.1876,
+      "step": 12469
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016968866749688665,
+      "loss": 0.1851,
+      "step": 12470
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0001696762141967621,
+      "loss": 0.1518,
+      "step": 12471
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0001696637608966376,
+      "loss": 0.1606,
+      "step": 12472
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016965130759651306,
+      "loss": 0.2652,
+      "step": 12473
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016963885429638854,
+      "loss": 0.3717,
+      "step": 12474
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016962640099626398,
+      "loss": 0.1869,
+      "step": 12475
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016961394769613946,
+      "loss": 0.4148,
+      "step": 12476
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016960149439601492,
+      "loss": 0.2269,
+      "step": 12477
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0001695890410958904,
+      "loss": 0.141,
+      "step": 12478
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016957658779576587,
+      "loss": 0.1873,
+      "step": 12479
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016956413449564135,
+      "loss": 0.2494,
+      "step": 12480
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016955168119551679,
+      "loss": 0.2242,
+      "step": 12481
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016953922789539225,
+      "loss": 0.2763,
+      "step": 12482
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00016952677459526773,
+      "loss": 0.2438,
+      "step": 12483
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0001695143212951432,
+      "loss": 0.1395,
+      "step": 12484
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016950186799501868,
+      "loss": 0.1569,
+      "step": 12485
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0001694894146948941,
+      "loss": 0.2552,
+      "step": 12486
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0001694769613947696,
+      "loss": 0.2486,
+      "step": 12487
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016946450809464505,
+      "loss": 0.3562,
+      "step": 12488
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016945205479452054,
+      "loss": 0.2053,
+      "step": 12489
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.000169439601494396,
+      "loss": 0.2238,
+      "step": 12490
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016942714819427149,
+      "loss": 0.271,
+      "step": 12491
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016941469489414692,
+      "loss": 0.3184,
+      "step": 12492
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0001694022415940224,
+      "loss": 0.3958,
+      "step": 12493
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016938978829389786,
+      "loss": 0.2429,
+      "step": 12494
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016937733499377335,
+      "loss": 0.1555,
+      "step": 12495
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0001693648816936488,
+      "loss": 0.2973,
+      "step": 12496
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016935242839352424,
+      "loss": 0.297,
+      "step": 12497
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016933997509339973,
+      "loss": 0.3085,
+      "step": 12498
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0001693275217932752,
+      "loss": 0.4223,
+      "step": 12499
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016931506849315067,
+      "loss": 0.2005,
+      "step": 12500
+    },
+    {
+      "epoch": 4.79,
+      "eval_cer": 0.10923386074097197,
+      "eval_loss": 0.33580130338668823,
+      "eval_runtime": 47.0502,
+      "eval_samples_per_second": 37.258,
+      "eval_steps_per_second": 4.676,
+      "eval_wer": 0.18998054034484319,
+      "step": 12500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016930261519302613,
+      "loss": 0.2224,
+      "step": 12501
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016929016189290162,
+      "loss": 0.2501,
+      "step": 12502
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016927770859277705,
+      "loss": 0.1884,
+      "step": 12503
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016926525529265254,
+      "loss": 0.2308,
+      "step": 12504
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.000169252801992528,
+      "loss": 0.329,
+      "step": 12505
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016924034869240348,
+      "loss": 0.3392,
+      "step": 12506
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016922789539227894,
+      "loss": 0.1848,
+      "step": 12507
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016921544209215443,
+      "loss": 0.146,
+      "step": 12508
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016920298879202986,
+      "loss": 0.1716,
+      "step": 12509
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00016919053549190535,
+      "loss": 0.2827,
+      "step": 12510
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0001691780821917808,
+      "loss": 0.247,
+      "step": 12511
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0001691656288916563,
+      "loss": 0.3475,
+      "step": 12512
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016915317559153175,
+      "loss": 0.2041,
+      "step": 12513
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016914072229140718,
+      "loss": 0.1922,
+      "step": 12514
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016912826899128267,
+      "loss": 0.1508,
+      "step": 12515
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016911581569115813,
+      "loss": 0.1136,
+      "step": 12516
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016910336239103361,
+      "loss": 0.2195,
+      "step": 12517
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016909090909090907,
+      "loss": 0.4125,
+      "step": 12518
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016907845579078456,
+      "loss": 0.3405,
+      "step": 12519
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016906600249066,
+      "loss": 0.1577,
+      "step": 12520
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016905354919053548,
+      "loss": 0.1556,
+      "step": 12521
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016904109589041094,
+      "loss": 0.3463,
+      "step": 12522
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016902864259028642,
+      "loss": 0.1945,
+      "step": 12523
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016901618929016188,
+      "loss": 0.353,
+      "step": 12524
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016900373599003737,
+      "loss": 0.206,
+      "step": 12525
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0001689912826899128,
+      "loss": 0.1811,
+      "step": 12526
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0001689788293897883,
+      "loss": 0.1724,
+      "step": 12527
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016896637608966375,
+      "loss": 0.2055,
+      "step": 12528
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0001689539227895392,
+      "loss": 0.2054,
+      "step": 12529
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0001689414694894147,
+      "loss": 0.291,
+      "step": 12530
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016892901618929012,
+      "loss": 0.2691,
+      "step": 12531
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0001689165628891656,
+      "loss": 0.2032,
+      "step": 12532
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016890410958904107,
+      "loss": 0.2175,
+      "step": 12533
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016889165628891656,
+      "loss": 0.1126,
+      "step": 12534
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00016887920298879202,
+      "loss": 0.1755,
+      "step": 12535
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0001688667496886675,
+      "loss": 0.332,
+      "step": 12536
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016885429638854293,
+      "loss": 0.1923,
+      "step": 12537
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016884184308841842,
+      "loss": 0.2119,
+      "step": 12538
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016882938978829388,
+      "loss": 0.1642,
+      "step": 12539
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016881693648816937,
+      "loss": 0.2891,
+      "step": 12540
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016880448318804482,
+      "loss": 0.1651,
+      "step": 12541
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0001687920298879203,
+      "loss": 0.2866,
+      "step": 12542
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016877957658779574,
+      "loss": 0.6067,
+      "step": 12543
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0001687671232876712,
+      "loss": 0.1686,
+      "step": 12544
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0001687546699875467,
+      "loss": 0.1759,
+      "step": 12545
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016874221668742215,
+      "loss": 0.1723,
+      "step": 12546
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016872976338729763,
+      "loss": 0.2341,
+      "step": 12547
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016871731008717307,
+      "loss": 0.2161,
+      "step": 12548
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016870485678704855,
+      "loss": 0.4922,
+      "step": 12549
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.000168692403486924,
+      "loss": 0.1812,
+      "step": 12550
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0001686799501867995,
+      "loss": 0.1959,
+      "step": 12551
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016866749688667496,
+      "loss": 0.1426,
+      "step": 12552
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016865504358655044,
+      "loss": 0.2719,
+      "step": 12553
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016864259028642587,
+      "loss": 0.1769,
+      "step": 12554
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016863013698630136,
+      "loss": 0.2969,
+      "step": 12555
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016861768368617682,
+      "loss": 0.2556,
+      "step": 12556
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0001686052303860523,
+      "loss": 0.1679,
+      "step": 12557
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016859277708592777,
+      "loss": 0.1516,
+      "step": 12558
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016858032378580325,
+      "loss": 0.1354,
+      "step": 12559
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016856787048567868,
+      "loss": 0.2744,
+      "step": 12560
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016855541718555414,
+      "loss": 0.3805,
+      "step": 12561
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00016854296388542963,
+      "loss": 0.2332,
+      "step": 12562
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0001685305105853051,
+      "loss": 0.3166,
+      "step": 12563
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016851805728518057,
+      "loss": 0.1781,
+      "step": 12564
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.000168505603985056,
+      "loss": 0.2253,
+      "step": 12565
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0001684931506849315,
+      "loss": 0.1611,
+      "step": 12566
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016848069738480695,
+      "loss": 0.3364,
+      "step": 12567
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016846824408468244,
+      "loss": 0.3387,
+      "step": 12568
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0001684557907845579,
+      "loss": 0.2425,
+      "step": 12569
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016844333748443338,
+      "loss": 0.3363,
+      "step": 12570
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016843088418430882,
+      "loss": 0.1887,
+      "step": 12571
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0001684184308841843,
+      "loss": 0.1715,
+      "step": 12572
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016840597758405976,
+      "loss": 0.2131,
+      "step": 12573
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016839352428393525,
+      "loss": 0.3122,
+      "step": 12574
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0001683810709838107,
+      "loss": 0.3507,
+      "step": 12575
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016836861768368614,
+      "loss": 0.1887,
+      "step": 12576
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016835616438356163,
+      "loss": 0.1668,
+      "step": 12577
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016834371108343708,
+      "loss": 0.1763,
+      "step": 12578
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016833125778331257,
+      "loss": 0.2829,
+      "step": 12579
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016831880448318803,
+      "loss": 0.3563,
+      "step": 12580
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016830635118306352,
+      "loss": 0.1609,
+      "step": 12581
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016829389788293895,
+      "loss": 0.267,
+      "step": 12582
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016828144458281443,
+      "loss": 0.1787,
+      "step": 12583
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0001682689912826899,
+      "loss": 0.1634,
+      "step": 12584
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016825653798256538,
+      "loss": 0.1505,
+      "step": 12585
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016824408468244084,
+      "loss": 0.3663,
+      "step": 12586
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016823163138231633,
+      "loss": 0.2882,
+      "step": 12587
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00016821917808219176,
+      "loss": 0.191,
+      "step": 12588
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016820672478206724,
+      "loss": 0.1664,
+      "step": 12589
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0001681942714819427,
+      "loss": 0.2927,
+      "step": 12590
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016818181818181816,
+      "loss": 0.1399,
+      "step": 12591
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016816936488169365,
+      "loss": 0.2175,
+      "step": 12592
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016815691158156908,
+      "loss": 0.2373,
+      "step": 12593
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016814445828144457,
+      "loss": 0.2075,
+      "step": 12594
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016813200498132003,
+      "loss": 0.163,
+      "step": 12595
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0001681195516811955,
+      "loss": 0.1818,
+      "step": 12596
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016810709838107097,
+      "loss": 0.1737,
+      "step": 12597
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016809464508094643,
+      "loss": 0.3519,
+      "step": 12598
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0001680821917808219,
+      "loss": 0.2388,
+      "step": 12599
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016806973848069738,
+      "loss": 0.2305,
+      "step": 12600
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016805728518057284,
+      "loss": 0.2151,
+      "step": 12601
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016804483188044832,
+      "loss": 0.1473,
+      "step": 12602
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016803237858032378,
+      "loss": 0.1397,
+      "step": 12603
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016801992528019924,
+      "loss": 0.1506,
+      "step": 12604
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0001680074719800747,
+      "loss": 0.4273,
+      "step": 12605
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016799501867995019,
+      "loss": 0.1866,
+      "step": 12606
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016798256537982564,
+      "loss": 0.1661,
+      "step": 12607
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0001679701120797011,
+      "loss": 0.2651,
+      "step": 12608
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016795765877957656,
+      "loss": 0.2492,
+      "step": 12609
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016794520547945202,
+      "loss": 0.2328,
+      "step": 12610
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0001679327521793275,
+      "loss": 0.2005,
+      "step": 12611
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016792029887920297,
+      "loss": 0.2054,
+      "step": 12612
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00016790784557907845,
+      "loss": 0.2399,
+      "step": 12613
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0001678953922789539,
+      "loss": 0.1569,
+      "step": 12614
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016788293897882937,
+      "loss": 0.2816,
+      "step": 12615
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016787048567870483,
+      "loss": 0.1454,
+      "step": 12616
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016785803237858032,
+      "loss": 0.2604,
+      "step": 12617
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016784557907845578,
+      "loss": 0.3019,
+      "step": 12618
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016783312577833126,
+      "loss": 0.1859,
+      "step": 12619
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0001678206724782067,
+      "loss": 0.2435,
+      "step": 12620
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016780821917808218,
+      "loss": 0.1893,
+      "step": 12621
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016779576587795764,
+      "loss": 0.2293,
+      "step": 12622
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0001677833125778331,
+      "loss": 0.3377,
+      "step": 12623
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016777085927770859,
+      "loss": 0.3341,
+      "step": 12624
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016775840597758404,
+      "loss": 0.1555,
+      "step": 12625
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0001677459526774595,
+      "loss": 0.1273,
+      "step": 12626
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016773349937733496,
+      "loss": 0.181,
+      "step": 12627
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016772104607721045,
+      "loss": 0.1654,
+      "step": 12628
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0001677085927770859,
+      "loss": 0.1949,
+      "step": 12629
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0001676961394769614,
+      "loss": 0.2836,
+      "step": 12630
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016768368617683683,
+      "loss": 0.2126,
+      "step": 12631
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0001676712328767123,
+      "loss": 0.1793,
+      "step": 12632
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016765877957658777,
+      "loss": 0.2323,
+      "step": 12633
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016764632627646326,
+      "loss": 0.3168,
+      "step": 12634
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016763387297633872,
+      "loss": 0.4823,
+      "step": 12635
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0001676214196762142,
+      "loss": 0.3157,
+      "step": 12636
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016760896637608964,
+      "loss": 0.1942,
+      "step": 12637
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0001675965130759651,
+      "loss": 0.2032,
+      "step": 12638
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016758405977584058,
+      "loss": 0.2708,
+      "step": 12639
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00016757160647571604,
+      "loss": 0.2442,
+      "step": 12640
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016755915317559153,
+      "loss": 0.1392,
+      "step": 12641
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016754669987546696,
+      "loss": 0.2084,
+      "step": 12642
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016753424657534245,
+      "loss": 0.2018,
+      "step": 12643
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001675217932752179,
+      "loss": 0.1708,
+      "step": 12644
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001675093399750934,
+      "loss": 0.2604,
+      "step": 12645
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016749688667496885,
+      "loss": 0.1701,
+      "step": 12646
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016748443337484434,
+      "loss": 0.1846,
+      "step": 12647
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016747198007471977,
+      "loss": 0.2217,
+      "step": 12648
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016745952677459525,
+      "loss": 0.2938,
+      "step": 12649
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016744707347447071,
+      "loss": 0.3784,
+      "step": 12650
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001674346201743462,
+      "loss": 0.2257,
+      "step": 12651
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016742216687422166,
+      "loss": 0.1724,
+      "step": 12652
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016740971357409715,
+      "loss": 0.2213,
+      "step": 12653
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016739726027397258,
+      "loss": 0.214,
+      "step": 12654
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016738480697384804,
+      "loss": 0.3444,
+      "step": 12655
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016737235367372352,
+      "loss": 0.2017,
+      "step": 12656
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016735990037359898,
+      "loss": 0.2862,
+      "step": 12657
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016734744707347447,
+      "loss": 0.1824,
+      "step": 12658
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001673349937733499,
+      "loss": 0.1699,
+      "step": 12659
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001673225404732254,
+      "loss": 0.2159,
+      "step": 12660
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016731008717310085,
+      "loss": 0.2882,
+      "step": 12661
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016729763387297633,
+      "loss": 0.3028,
+      "step": 12662
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001672851805728518,
+      "loss": 0.2286,
+      "step": 12663
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00016727272727272728,
+      "loss": 0.2495,
+      "step": 12664
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001672602739726027,
+      "loss": 0.1914,
+      "step": 12665
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001672478206724782,
+      "loss": 0.2557,
+      "step": 12666
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016723536737235366,
+      "loss": 0.3483,
+      "step": 12667
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016722291407222914,
+      "loss": 0.3156,
+      "step": 12668
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001672104607721046,
+      "loss": 0.2155,
+      "step": 12669
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016719800747198003,
+      "loss": 0.2782,
+      "step": 12670
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016718555417185552,
+      "loss": 0.1902,
+      "step": 12671
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016717310087173098,
+      "loss": 0.1742,
+      "step": 12672
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016716064757160646,
+      "loss": 0.1973,
+      "step": 12673
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016714819427148192,
+      "loss": 0.3334,
+      "step": 12674
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001671357409713574,
+      "loss": 0.2293,
+      "step": 12675
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016712328767123284,
+      "loss": 0.2721,
+      "step": 12676
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016711083437110833,
+      "loss": 0.3196,
+      "step": 12677
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001670983810709838,
+      "loss": 0.1717,
+      "step": 12678
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016708592777085927,
+      "loss": 0.1835,
+      "step": 12679
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016707347447073473,
+      "loss": 0.2541,
+      "step": 12680
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016706102117061022,
+      "loss": 0.2043,
+      "step": 12681
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016704856787048565,
+      "loss": 0.2534,
+      "step": 12682
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016703611457036114,
+      "loss": 0.1615,
+      "step": 12683
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001670236612702366,
+      "loss": 0.247,
+      "step": 12684
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016701120797011206,
+      "loss": 0.1572,
+      "step": 12685
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016699875466998754,
+      "loss": 0.2627,
+      "step": 12686
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016698630136986297,
+      "loss": 0.1831,
+      "step": 12687
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016697384806973846,
+      "loss": 0.2365,
+      "step": 12688
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016696139476961392,
+      "loss": 0.1566,
+      "step": 12689
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001669489414694894,
+      "loss": 0.202,
+      "step": 12690
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016693648816936487,
+      "loss": 0.2283,
+      "step": 12691
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00016692403486924035,
+      "loss": 0.1897,
+      "step": 12692
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016691158156911578,
+      "loss": 0.405,
+      "step": 12693
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016689912826899127,
+      "loss": 0.2193,
+      "step": 12694
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016688667496886673,
+      "loss": 0.3132,
+      "step": 12695
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016687422166874222,
+      "loss": 0.139,
+      "step": 12696
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016686176836861767,
+      "loss": 0.1824,
+      "step": 12697
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016684931506849316,
+      "loss": 0.2952,
+      "step": 12698
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001668368617683686,
+      "loss": 0.2531,
+      "step": 12699
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016682440846824408,
+      "loss": 0.2111,
+      "step": 12700
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016681195516811954,
+      "loss": 0.2699,
+      "step": 12701
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.000166799501867995,
+      "loss": 0.1821,
+      "step": 12702
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016678704856787048,
+      "loss": 0.165,
+      "step": 12703
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016677459526774592,
+      "loss": 0.3172,
+      "step": 12704
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001667621419676214,
+      "loss": 0.4368,
+      "step": 12705
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016674968866749686,
+      "loss": 0.2152,
+      "step": 12706
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016673723536737235,
+      "loss": 0.1666,
+      "step": 12707
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001667247820672478,
+      "loss": 0.1629,
+      "step": 12708
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001667123287671233,
+      "loss": 0.2513,
+      "step": 12709
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016669987546699872,
+      "loss": 0.263,
+      "step": 12710
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001666874221668742,
+      "loss": 0.366,
+      "step": 12711
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016667496886674967,
+      "loss": 0.2829,
+      "step": 12712
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016666251556662516,
+      "loss": 0.1621,
+      "step": 12713
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016665006226650062,
+      "loss": 0.1779,
+      "step": 12714
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001666376089663761,
+      "loss": 0.1511,
+      "step": 12715
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016662515566625153,
+      "loss": 0.1862,
+      "step": 12716
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.000166612702366127,
+      "loss": 0.2842,
+      "step": 12717
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00016660024906600248,
+      "loss": 0.3922,
+      "step": 12718
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016658779576587794,
+      "loss": 0.1877,
+      "step": 12719
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016657534246575342,
+      "loss": 0.2418,
+      "step": 12720
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016656288916562886,
+      "loss": 0.2286,
+      "step": 12721
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016655043586550434,
+      "loss": 0.1662,
+      "step": 12722
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001665379825653798,
+      "loss": 0.1968,
+      "step": 12723
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001665255292652553,
+      "loss": 0.5103,
+      "step": 12724
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016651307596513075,
+      "loss": 0.2954,
+      "step": 12725
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016650062266500623,
+      "loss": 0.2349,
+      "step": 12726
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016648816936488167,
+      "loss": 0.2154,
+      "step": 12727
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016647571606475715,
+      "loss": 0.1936,
+      "step": 12728
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001664632627646326,
+      "loss": 0.187,
+      "step": 12729
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001664508094645081,
+      "loss": 0.4345,
+      "step": 12730
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016643835616438356,
+      "loss": 0.1933,
+      "step": 12731
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.000166425902864259,
+      "loss": 0.2365,
+      "step": 12732
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016641344956413448,
+      "loss": 0.2142,
+      "step": 12733
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016640099626400993,
+      "loss": 0.1232,
+      "step": 12734
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016638854296388542,
+      "loss": 0.2281,
+      "step": 12735
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016637608966376088,
+      "loss": 0.2884,
+      "step": 12736
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016636363636363637,
+      "loss": 0.1929,
+      "step": 12737
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001663511830635118,
+      "loss": 0.2158,
+      "step": 12738
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016633872976338728,
+      "loss": 0.1819,
+      "step": 12739
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016632627646326274,
+      "loss": 0.1576,
+      "step": 12740
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016631382316313823,
+      "loss": 0.1264,
+      "step": 12741
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001663013698630137,
+      "loss": 0.345,
+      "step": 12742
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00016628891656288918,
+      "loss": 0.2651,
+      "step": 12743
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001662764632627646,
+      "loss": 0.2316,
+      "step": 12744
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0001662640099626401,
+      "loss": 0.2312,
+      "step": 12745
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016625155666251555,
+      "loss": 0.1927,
+      "step": 12746
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016623910336239104,
+      "loss": 0.1543,
+      "step": 12747
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0001662266500622665,
+      "loss": 0.2015,
+      "step": 12748
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016621419676214193,
+      "loss": 0.3311,
+      "step": 12749
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016620174346201742,
+      "loss": 0.2676,
+      "step": 12750
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016618929016189288,
+      "loss": 0.1473,
+      "step": 12751
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016617683686176836,
+      "loss": 0.128,
+      "step": 12752
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016616438356164382,
+      "loss": 0.2495,
+      "step": 12753
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0001661519302615193,
+      "loss": 0.2209,
+      "step": 12754
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016613947696139474,
+      "loss": 0.3328,
+      "step": 12755
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016612702366127023,
+      "loss": 0.161,
+      "step": 12756
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016611457036114569,
+      "loss": 0.1542,
+      "step": 12757
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016610211706102117,
+      "loss": 0.2151,
+      "step": 12758
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016608966376089663,
+      "loss": 0.098,
+      "step": 12759
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0001660772104607721,
+      "loss": 0.1092,
+      "step": 12760
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016606475716064755,
+      "loss": 0.4189,
+      "step": 12761
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016605230386052304,
+      "loss": 0.2265,
+      "step": 12762
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0001660398505603985,
+      "loss": 0.3015,
+      "step": 12763
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016602739726027395,
+      "loss": 0.1148,
+      "step": 12764
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016601494396014944,
+      "loss": 0.134,
+      "step": 12765
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016600249066002487,
+      "loss": 0.25,
+      "step": 12766
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016599003735990036,
+      "loss": 0.2558,
+      "step": 12767
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016597758405977582,
+      "loss": 0.3022,
+      "step": 12768
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0001659651307596513,
+      "loss": 0.1928,
+      "step": 12769
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016595267745952676,
+      "loss": 0.1981,
+      "step": 12770
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00016594022415940222,
+      "loss": 0.2229,
+      "step": 12771
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016592777085927768,
+      "loss": 0.3233,
+      "step": 12772
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016591531755915317,
+      "loss": 0.1374,
+      "step": 12773
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016590286425902863,
+      "loss": 0.2775,
+      "step": 12774
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0001658904109589041,
+      "loss": 0.2156,
+      "step": 12775
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016587795765877957,
+      "loss": 0.169,
+      "step": 12776
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016586550435865503,
+      "loss": 0.1418,
+      "step": 12777
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0001658530510585305,
+      "loss": 0.1994,
+      "step": 12778
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016584059775840595,
+      "loss": 0.1986,
+      "step": 12779
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016582814445828144,
+      "loss": 0.1973,
+      "step": 12780
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0001658156911581569,
+      "loss": 0.2105,
+      "step": 12781
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016580323785803235,
+      "loss": 0.2897,
+      "step": 12782
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0001657907845579078,
+      "loss": 0.2031,
+      "step": 12783
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0001657783312577833,
+      "loss": 0.1885,
+      "step": 12784
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016576587795765876,
+      "loss": 0.3234,
+      "step": 12785
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016575342465753425,
+      "loss": 0.2596,
+      "step": 12786
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0001657409713574097,
+      "loss": 0.2135,
+      "step": 12787
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016572851805728516,
+      "loss": 0.1409,
+      "step": 12788
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016571606475716062,
+      "loss": 0.1995,
+      "step": 12789
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0001657036114570361,
+      "loss": 0.155,
+      "step": 12790
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016569115815691157,
+      "loss": 0.1493,
+      "step": 12791
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016567870485678705,
+      "loss": 0.1544,
+      "step": 12792
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016566625155666249,
+      "loss": 0.4063,
+      "step": 12793
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016565379825653797,
+      "loss": 0.2966,
+      "step": 12794
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016564134495641343,
+      "loss": 0.3209,
+      "step": 12795
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0001656288916562889,
+      "loss": 0.3412,
+      "step": 12796
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00016561643835616438,
+      "loss": 0.1403,
+      "step": 12797
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016560398505603984,
+      "loss": 0.2569,
+      "step": 12798
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0001655915317559153,
+      "loss": 0.3365,
+      "step": 12799
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016557907845579075,
+      "loss": 0.2048,
+      "step": 12800
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016556662515566624,
+      "loss": 0.1861,
+      "step": 12801
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0001655541718555417,
+      "loss": 0.1728,
+      "step": 12802
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016554171855541719,
+      "loss": 0.1346,
+      "step": 12803
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016552926525529262,
+      "loss": 0.2424,
+      "step": 12804
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0001655168119551681,
+      "loss": 0.3058,
+      "step": 12805
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016550435865504356,
+      "loss": 0.2578,
+      "step": 12806
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016549190535491905,
+      "loss": 0.1626,
+      "step": 12807
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0001654794520547945,
+      "loss": 0.154,
+      "step": 12808
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016546699875467,
+      "loss": 0.1948,
+      "step": 12809
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016545454545454543,
+      "loss": 0.1201,
+      "step": 12810
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0001654420921544209,
+      "loss": 0.3344,
+      "step": 12811
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016542963885429637,
+      "loss": 0.2929,
+      "step": 12812
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016541718555417183,
+      "loss": 0.2629,
+      "step": 12813
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016540473225404732,
+      "loss": 0.2725,
+      "step": 12814
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016539227895392275,
+      "loss": 0.1164,
+      "step": 12815
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016537982565379824,
+      "loss": 0.1374,
+      "step": 12816
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0001653673723536737,
+      "loss": 0.1987,
+      "step": 12817
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016535491905354918,
+      "loss": 0.2736,
+      "step": 12818
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016534246575342464,
+      "loss": 0.2027,
+      "step": 12819
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016533001245330013,
+      "loss": 0.2144,
+      "step": 12820
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016531755915317556,
+      "loss": 0.1851,
+      "step": 12821
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00016530510585305105,
+      "loss": 0.1685,
+      "step": 12822
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0001652926525529265,
+      "loss": 0.2976,
+      "step": 12823
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.000165280199252802,
+      "loss": 0.3678,
+      "step": 12824
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016526774595267745,
+      "loss": 0.1576,
+      "step": 12825
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016525529265255294,
+      "loss": 0.149,
+      "step": 12826
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016524283935242837,
+      "loss": 0.1727,
+      "step": 12827
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016523038605230383,
+      "loss": 0.1502,
+      "step": 12828
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016521793275217931,
+      "loss": 0.1982,
+      "step": 12829
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016520547945205477,
+      "loss": 0.2417,
+      "step": 12830
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016519302615193026,
+      "loss": 0.1847,
+      "step": 12831
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0001651805728518057,
+      "loss": 0.3221,
+      "step": 12832
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016516811955168118,
+      "loss": 0.1275,
+      "step": 12833
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016515566625155664,
+      "loss": 0.1522,
+      "step": 12834
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016514321295143212,
+      "loss": 0.1447,
+      "step": 12835
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016513075965130758,
+      "loss": 0.4395,
+      "step": 12836
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016511830635118307,
+      "loss": 0.1763,
+      "step": 12837
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0001651058530510585,
+      "loss": 0.2741,
+      "step": 12838
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.000165093399750934,
+      "loss": 0.1701,
+      "step": 12839
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016508094645080945,
+      "loss": 0.1762,
+      "step": 12840
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016506849315068493,
+      "loss": 0.1853,
+      "step": 12841
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0001650560398505604,
+      "loss": 0.2736,
+      "step": 12842
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016504358655043582,
+      "loss": 0.2755,
+      "step": 12843
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0001650311332503113,
+      "loss": 0.1714,
+      "step": 12844
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016501867995018677,
+      "loss": 0.2195,
+      "step": 12845
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016500622665006226,
+      "loss": 0.1495,
+      "step": 12846
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016499377334993772,
+      "loss": 0.1838,
+      "step": 12847
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0001649813200498132,
+      "loss": 0.3044,
+      "step": 12848
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00016496886674968863,
+      "loss": 0.3339,
+      "step": 12849
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016495641344956412,
+      "loss": 0.4213,
+      "step": 12850
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016494396014943958,
+      "loss": 0.1455,
+      "step": 12851
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016493150684931507,
+      "loss": 0.4083,
+      "step": 12852
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016491905354919052,
+      "loss": 0.1915,
+      "step": 12853
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.000164906600249066,
+      "loss": 0.1049,
+      "step": 12854
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016489414694894144,
+      "loss": 0.295,
+      "step": 12855
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016488169364881693,
+      "loss": 0.2471,
+      "step": 12856
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0001648692403486924,
+      "loss": 0.1785,
+      "step": 12857
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016485678704856785,
+      "loss": 0.1638,
+      "step": 12858
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016484433374844333,
+      "loss": 0.2038,
+      "step": 12859
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016483188044831877,
+      "loss": 0.1857,
+      "step": 12860
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016481942714819425,
+      "loss": 0.2117,
+      "step": 12861
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0001648069738480697,
+      "loss": 0.2033,
+      "step": 12862
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0001647945205479452,
+      "loss": 0.2821,
+      "step": 12863
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016478206724782066,
+      "loss": 0.2146,
+      "step": 12864
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016476961394769614,
+      "loss": 0.2103,
+      "step": 12865
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016475716064757157,
+      "loss": 0.1631,
+      "step": 12866
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016474470734744706,
+      "loss": 0.2628,
+      "step": 12867
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016473225404732252,
+      "loss": 0.2386,
+      "step": 12868
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.000164719800747198,
+      "loss": 0.2116,
+      "step": 12869
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016470734744707347,
+      "loss": 0.1661,
+      "step": 12870
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016469489414694895,
+      "loss": 0.2664,
+      "step": 12871
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016468244084682438,
+      "loss": 0.1519,
+      "step": 12872
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016466998754669987,
+      "loss": 0.2564,
+      "step": 12873
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00016465753424657533,
+      "loss": 0.2127,
+      "step": 12874
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0001646450809464508,
+      "loss": 0.1896,
+      "step": 12875
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016463262764632627,
+      "loss": 0.165,
+      "step": 12876
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0001646201743462017,
+      "loss": 0.1381,
+      "step": 12877
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0001646077210460772,
+      "loss": 0.1925,
+      "step": 12878
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016459526774595265,
+      "loss": 0.1713,
+      "step": 12879
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016458281444582814,
+      "loss": 0.2627,
+      "step": 12880
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0001645703611457036,
+      "loss": 0.2463,
+      "step": 12881
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016455790784557908,
+      "loss": 0.1784,
+      "step": 12882
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016454545454545452,
+      "loss": 0.1495,
+      "step": 12883
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016453300124533,
+      "loss": 0.1775,
+      "step": 12884
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016452054794520546,
+      "loss": 0.1871,
+      "step": 12885
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016450809464508095,
+      "loss": 0.256,
+      "step": 12886
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0001644956413449564,
+      "loss": 0.2087,
+      "step": 12887
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0001644831880448319,
+      "loss": 0.1874,
+      "step": 12888
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016447073474470733,
+      "loss": 0.1562,
+      "step": 12889
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016445828144458278,
+      "loss": 0.1358,
+      "step": 12890
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016444582814445827,
+      "loss": 0.2478,
+      "step": 12891
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016443337484433373,
+      "loss": 0.1553,
+      "step": 12892
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016442092154420922,
+      "loss": 0.1887,
+      "step": 12893
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016440846824408465,
+      "loss": 0.3874,
+      "step": 12894
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016439601494396013,
+      "loss": 0.1733,
+      "step": 12895
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0001643835616438356,
+      "loss": 0.1514,
+      "step": 12896
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016437110834371108,
+      "loss": 0.2798,
+      "step": 12897
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016435865504358654,
+      "loss": 0.1892,
+      "step": 12898
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016434620174346203,
+      "loss": 0.3105,
+      "step": 12899
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016433374844333746,
+      "loss": 0.2641,
+      "step": 12900
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00016432129514321294,
+      "loss": 0.2194,
+      "step": 12901
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0001643088418430884,
+      "loss": 0.1602,
+      "step": 12902
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0001642963885429639,
+      "loss": 0.148,
+      "step": 12903
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016428393524283935,
+      "loss": 0.2467,
+      "step": 12904
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016427148194271478,
+      "loss": 0.347,
+      "step": 12905
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016425902864259027,
+      "loss": 0.3436,
+      "step": 12906
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016424657534246573,
+      "loss": 0.1971,
+      "step": 12907
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0001642341220423412,
+      "loss": 0.211,
+      "step": 12908
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016422166874221667,
+      "loss": 0.2235,
+      "step": 12909
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016420921544209216,
+      "loss": 0.2404,
+      "step": 12910
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0001641967621419676,
+      "loss": 0.2643,
+      "step": 12911
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016418430884184308,
+      "loss": 0.1944,
+      "step": 12912
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016417185554171854,
+      "loss": 0.1878,
+      "step": 12913
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016415940224159402,
+      "loss": 0.1759,
+      "step": 12914
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016414694894146948,
+      "loss": 0.1776,
+      "step": 12915
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016413449564134497,
+      "loss": 0.1602,
+      "step": 12916
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0001641220423412204,
+      "loss": 0.2617,
+      "step": 12917
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016410958904109589,
+      "loss": 0.4211,
+      "step": 12918
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016409713574097134,
+      "loss": 0.1783,
+      "step": 12919
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016408468244084683,
+      "loss": 0.1494,
+      "step": 12920
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0001640722291407223,
+      "loss": 0.2102,
+      "step": 12921
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016405977584059772,
+      "loss": 0.1321,
+      "step": 12922
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0001640473225404732,
+      "loss": 0.1916,
+      "step": 12923
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016403486924034867,
+      "loss": 0.2551,
+      "step": 12924
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00016402241594022415,
+      "loss": 0.2108,
+      "step": 12925
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0001640099626400996,
+      "loss": 0.1996,
+      "step": 12926
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0001639975093399751,
+      "loss": 0.3736,
+      "step": 12927
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016398505603985053,
+      "loss": 0.15,
+      "step": 12928
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016397260273972602,
+      "loss": 0.1346,
+      "step": 12929
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016396014943960148,
+      "loss": 0.2525,
+      "step": 12930
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016394769613947696,
+      "loss": 0.2464,
+      "step": 12931
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016393524283935242,
+      "loss": 0.1571,
+      "step": 12932
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016392278953922788,
+      "loss": 0.1813,
+      "step": 12933
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016391033623910334,
+      "loss": 0.155,
+      "step": 12934
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016389788293897883,
+      "loss": 0.2568,
+      "step": 12935
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016388542963885429,
+      "loss": 0.2565,
+      "step": 12936
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016387297633872975,
+      "loss": 0.2659,
+      "step": 12937
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016386052303860523,
+      "loss": 0.2623,
+      "step": 12938
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016384806973848066,
+      "loss": 0.1803,
+      "step": 12939
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016383561643835615,
+      "loss": 0.193,
+      "step": 12940
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0001638231631382316,
+      "loss": 0.249,
+      "step": 12941
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0001638107098381071,
+      "loss": 0.439,
+      "step": 12942
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016379825653798255,
+      "loss": 0.2682,
+      "step": 12943
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.000163785803237858,
+      "loss": 0.2769,
+      "step": 12944
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016377334993773347,
+      "loss": 0.1553,
+      "step": 12945
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016376089663760896,
+      "loss": 0.2046,
+      "step": 12946
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016374844333748442,
+      "loss": 0.1962,
+      "step": 12947
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0001637359900373599,
+      "loss": 0.2341,
+      "step": 12948
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016372353673723536,
+      "loss": 0.3425,
+      "step": 12949
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016371108343711082,
+      "loss": 0.2937,
+      "step": 12950
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016369863013698628,
+      "loss": 0.1711,
+      "step": 12951
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016368617683686174,
+      "loss": 0.1592,
+      "step": 12952
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00016367372353673723,
+      "loss": 0.2174,
+      "step": 12953
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016366127023661269,
+      "loss": 0.2499,
+      "step": 12954
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016364881693648815,
+      "loss": 0.2566,
+      "step": 12955
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0001636363636363636,
+      "loss": 0.2283,
+      "step": 12956
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0001636239103362391,
+      "loss": 0.1292,
+      "step": 12957
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016361145703611455,
+      "loss": 0.2093,
+      "step": 12958
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016359900373599004,
+      "loss": 0.1335,
+      "step": 12959
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0001635865504358655,
+      "loss": 0.1996,
+      "step": 12960
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016357409713574095,
+      "loss": 0.4287,
+      "step": 12961
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016356164383561641,
+      "loss": 0.2394,
+      "step": 12962
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0001635491905354919,
+      "loss": 0.2459,
+      "step": 12963
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016353673723536736,
+      "loss": 0.2263,
+      "step": 12964
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016352428393524285,
+      "loss": 0.143,
+      "step": 12965
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016351183063511828,
+      "loss": 0.1729,
+      "step": 12966
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016349937733499376,
+      "loss": 0.1575,
+      "step": 12967
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016348692403486922,
+      "loss": 0.3667,
+      "step": 12968
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016347447073474468,
+      "loss": 0.1608,
+      "step": 12969
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016346201743462017,
+      "loss": 0.174,
+      "step": 12970
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0001634495641344956,
+      "loss": 0.1991,
+      "step": 12971
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0001634371108343711,
+      "loss": 0.2081,
+      "step": 12972
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016342465753424655,
+      "loss": 0.2736,
+      "step": 12973
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016341220423412203,
+      "loss": 0.2752,
+      "step": 12974
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0001633997509339975,
+      "loss": 0.1811,
+      "step": 12975
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016338729763387298,
+      "loss": 0.2178,
+      "step": 12976
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0001633748443337484,
+      "loss": 0.2075,
+      "step": 12977
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0001633623910336239,
+      "loss": 0.1851,
+      "step": 12978
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00016334993773349936,
+      "loss": 0.213,
+      "step": 12979
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016333748443337484,
+      "loss": 0.3053,
+      "step": 12980
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0001633250311332503,
+      "loss": 0.2513,
+      "step": 12981
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0001633125778331258,
+      "loss": 0.1798,
+      "step": 12982
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016330012453300122,
+      "loss": 0.1955,
+      "step": 12983
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016328767123287668,
+      "loss": 0.1815,
+      "step": 12984
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016327521793275216,
+      "loss": 0.2083,
+      "step": 12985
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016326276463262762,
+      "loss": 0.3255,
+      "step": 12986
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0001632503113325031,
+      "loss": 0.1697,
+      "step": 12987
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016323785803237854,
+      "loss": 0.2283,
+      "step": 12988
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016322540473225403,
+      "loss": 0.2433,
+      "step": 12989
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0001632129514321295,
+      "loss": 0.1505,
+      "step": 12990
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016320049813200497,
+      "loss": 0.2043,
+      "step": 12991
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016318804483188043,
+      "loss": 0.2369,
+      "step": 12992
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016317559153175592,
+      "loss": 0.2941,
+      "step": 12993
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016316313823163135,
+      "loss": 0.3764,
+      "step": 12994
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016315068493150684,
+      "loss": 0.283,
+      "step": 12995
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0001631382316313823,
+      "loss": 0.1903,
+      "step": 12996
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016312577833125778,
+      "loss": 0.1587,
+      "step": 12997
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016311332503113324,
+      "loss": 0.1492,
+      "step": 12998
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016310087173100867,
+      "loss": 0.2642,
+      "step": 12999
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016308841843088416,
+      "loss": 0.2982,
+      "step": 13000
+    },
+    {
+      "epoch": 4.98,
+      "eval_cer": 0.1060421015666153,
+      "eval_loss": 0.35133054852485657,
+      "eval_runtime": 47.1831,
+      "eval_samples_per_second": 37.153,
+      "eval_steps_per_second": 4.663,
+      "eval_wer": 0.17663031180703262,
+      "step": 13000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016307596513075962,
+      "loss": 0.2001,
+      "step": 13001
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0001630635118306351,
+      "loss": 0.2143,
+      "step": 13002
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016305105853051057,
+      "loss": 0.1942,
+      "step": 13003
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016303860523038605,
+      "loss": 0.1977,
+      "step": 13004
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00016302615193026148,
+      "loss": 0.2466,
+      "step": 13005
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016301369863013697,
+      "loss": 0.3192,
+      "step": 13006
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016300124533001243,
+      "loss": 0.1784,
+      "step": 13007
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016298879202988792,
+      "loss": 0.175,
+      "step": 13008
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016297633872976337,
+      "loss": 0.1894,
+      "step": 13009
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016296388542963886,
+      "loss": 0.1727,
+      "step": 13010
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001629514321295143,
+      "loss": 0.2607,
+      "step": 13011
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016293897882938978,
+      "loss": 0.2702,
+      "step": 13012
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016292652552926524,
+      "loss": 0.2062,
+      "step": 13013
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016291407222914072,
+      "loss": 0.1971,
+      "step": 13014
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016290161892901618,
+      "loss": 0.1782,
+      "step": 13015
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016288916562889162,
+      "loss": 0.158,
+      "step": 13016
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001628767123287671,
+      "loss": 0.3105,
+      "step": 13017
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016286425902864256,
+      "loss": 0.3488,
+      "step": 13018
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016285180572851805,
+      "loss": 0.2618,
+      "step": 13019
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001628393524283935,
+      "loss": 0.2106,
+      "step": 13020
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.000162826899128269,
+      "loss": 0.2323,
+      "step": 13021
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016281444582814442,
+      "loss": 0.296,
+      "step": 13022
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001628019925280199,
+      "loss": 0.3209,
+      "step": 13023
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016278953922789537,
+      "loss": 0.3574,
+      "step": 13024
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016277708592777086,
+      "loss": 0.2993,
+      "step": 13025
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016276463262764632,
+      "loss": 0.2242,
+      "step": 13026
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001627521793275218,
+      "loss": 0.3352,
+      "step": 13027
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016273972602739723,
+      "loss": 0.1736,
+      "step": 13028
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016272727272727272,
+      "loss": 0.2113,
+      "step": 13029
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016271481942714818,
+      "loss": 0.3486,
+      "step": 13030
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00016270236612702364,
+      "loss": 0.215,
+      "step": 13031
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016268991282689913,
+      "loss": 0.1859,
+      "step": 13032
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016267745952677456,
+      "loss": 0.1992,
+      "step": 13033
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016266500622665004,
+      "loss": 0.2018,
+      "step": 13034
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0001626525529265255,
+      "loss": 0.2574,
+      "step": 13035
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.000162640099626401,
+      "loss": 0.2551,
+      "step": 13036
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016262764632627645,
+      "loss": 0.3087,
+      "step": 13037
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016261519302615193,
+      "loss": 0.1999,
+      "step": 13038
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016260273972602737,
+      "loss": 0.2434,
+      "step": 13039
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016259028642590285,
+      "loss": 0.1649,
+      "step": 13040
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0001625778331257783,
+      "loss": 0.2062,
+      "step": 13041
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0001625653798256538,
+      "loss": 0.1705,
+      "step": 13042
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016255292652552926,
+      "loss": 0.3259,
+      "step": 13043
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016254047322540474,
+      "loss": 0.1946,
+      "step": 13044
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016252801992528018,
+      "loss": 0.2311,
+      "step": 13045
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016251556662515563,
+      "loss": 0.2677,
+      "step": 13046
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016250311332503112,
+      "loss": 0.1723,
+      "step": 13047
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016249066002490658,
+      "loss": 0.1826,
+      "step": 13048
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016247820672478207,
+      "loss": 0.2091,
+      "step": 13049
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0001624657534246575,
+      "loss": 0.25,
+      "step": 13050
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016245330012453298,
+      "loss": 0.2343,
+      "step": 13051
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016244084682440844,
+      "loss": 0.3316,
+      "step": 13052
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016242839352428393,
+      "loss": 0.1312,
+      "step": 13053
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0001624159402241594,
+      "loss": 0.1619,
+      "step": 13054
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016240348692403488,
+      "loss": 0.1939,
+      "step": 13055
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0001623910336239103,
+      "loss": 0.1835,
+      "step": 13056
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0001623785803237858,
+      "loss": 0.1412,
+      "step": 13057
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00016236612702366125,
+      "loss": 0.4038,
+      "step": 13058
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016235367372353674,
+      "loss": 0.1647,
+      "step": 13059
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0001623412204234122,
+      "loss": 0.3436,
+      "step": 13060
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016232876712328768,
+      "loss": 0.3604,
+      "step": 13061
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016231631382316312,
+      "loss": 0.2151,
+      "step": 13062
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016230386052303858,
+      "loss": 0.2069,
+      "step": 13063
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016229140722291406,
+      "loss": 0.3485,
+      "step": 13064
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016227895392278952,
+      "loss": 0.2184,
+      "step": 13065
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.000162266500622665,
+      "loss": 0.166,
+      "step": 13066
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016225404732254044,
+      "loss": 0.2104,
+      "step": 13067
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016224159402241593,
+      "loss": 0.2212,
+      "step": 13068
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016222914072229139,
+      "loss": 0.1726,
+      "step": 13069
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016221668742216687,
+      "loss": 0.334,
+      "step": 13070
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016220423412204233,
+      "loss": 0.1891,
+      "step": 13071
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016219178082191782,
+      "loss": 0.1907,
+      "step": 13072
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016217932752179325,
+      "loss": 0.262,
+      "step": 13073
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016216687422166874,
+      "loss": 0.1056,
+      "step": 13074
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0001621544209215442,
+      "loss": 0.1931,
+      "step": 13075
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016214196762141968,
+      "loss": 0.181,
+      "step": 13076
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016212951432129514,
+      "loss": 0.3156,
+      "step": 13077
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016211706102117057,
+      "loss": 0.1749,
+      "step": 13078
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016210460772104606,
+      "loss": 0.2051,
+      "step": 13079
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016209215442092152,
+      "loss": 0.1487,
+      "step": 13080
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.000162079701120797,
+      "loss": 0.1574,
+      "step": 13081
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016206724782067246,
+      "loss": 0.2826,
+      "step": 13082
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016205479452054795,
+      "loss": 0.319,
+      "step": 13083
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00016204234122042338,
+      "loss": 0.1952,
+      "step": 13084
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016202988792029887,
+      "loss": 0.1894,
+      "step": 13085
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016201743462017433,
+      "loss": 0.1507,
+      "step": 13086
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0001620049813200498,
+      "loss": 0.213,
+      "step": 13087
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016199252801992527,
+      "loss": 0.1755,
+      "step": 13088
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016198007471980076,
+      "loss": 0.299,
+      "step": 13089
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0001619676214196762,
+      "loss": 0.2848,
+      "step": 13090
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016195516811955168,
+      "loss": 0.1412,
+      "step": 13091
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016194271481942714,
+      "loss": 0.1777,
+      "step": 13092
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016193026151930262,
+      "loss": 0.1337,
+      "step": 13093
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016191780821917808,
+      "loss": 0.2381,
+      "step": 13094
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0001619053549190535,
+      "loss": 0.2203,
+      "step": 13095
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.000161892901618929,
+      "loss": 0.2121,
+      "step": 13096
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016188044831880446,
+      "loss": 0.2323,
+      "step": 13097
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016186799501867995,
+      "loss": 0.1674,
+      "step": 13098
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0001618555417185554,
+      "loss": 0.1281,
+      "step": 13099
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0001618430884184309,
+      "loss": 0.1238,
+      "step": 13100
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016183063511830632,
+      "loss": 0.1077,
+      "step": 13101
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0001618181818181818,
+      "loss": 0.2927,
+      "step": 13102
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016180572851805727,
+      "loss": 0.2055,
+      "step": 13103
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016179327521793275,
+      "loss": 0.1448,
+      "step": 13104
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0001617808219178082,
+      "loss": 0.2942,
+      "step": 13105
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016176836861768367,
+      "loss": 0.213,
+      "step": 13106
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016175591531755913,
+      "loss": 0.1273,
+      "step": 13107
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016174346201743462,
+      "loss": 0.2222,
+      "step": 13108
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016173100871731008,
+      "loss": 0.2235,
+      "step": 13109
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00016171855541718554,
+      "loss": 0.1778,
+      "step": 13110
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016170610211706102,
+      "loss": 0.2001,
+      "step": 13111
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016169364881693645,
+      "loss": 0.1047,
+      "step": 13112
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016168119551681194,
+      "loss": 0.2322,
+      "step": 13113
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0001616687422166874,
+      "loss": 0.2648,
+      "step": 13114
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0001616562889165629,
+      "loss": 0.1947,
+      "step": 13115
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016164383561643835,
+      "loss": 0.1371,
+      "step": 13116
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0001616313823163138,
+      "loss": 0.1561,
+      "step": 13117
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016161892901618926,
+      "loss": 0.1565,
+      "step": 13118
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016160647571606475,
+      "loss": 0.1733,
+      "step": 13119
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0001615940224159402,
+      "loss": 0.323,
+      "step": 13120
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0001615815691158157,
+      "loss": 0.1524,
+      "step": 13121
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016156911581569115,
+      "loss": 0.2119,
+      "step": 13122
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016155666251556661,
+      "loss": 0.2055,
+      "step": 13123
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016154420921544207,
+      "loss": 0.1632,
+      "step": 13124
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016153175591531753,
+      "loss": 0.1573,
+      "step": 13125
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016151930261519302,
+      "loss": 0.2513,
+      "step": 13126
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016150684931506848,
+      "loss": 0.2722,
+      "step": 13127
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016149439601494394,
+      "loss": 0.2271,
+      "step": 13128
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0001614819427148194,
+      "loss": 0.1557,
+      "step": 13129
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016146948941469488,
+      "loss": 0.1824,
+      "step": 13130
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016145703611457034,
+      "loss": 0.1709,
+      "step": 13131
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016144458281444583,
+      "loss": 0.2427,
+      "step": 13132
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016143212951432126,
+      "loss": 0.313,
+      "step": 13133
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.00016141967621419675,
+      "loss": 0.2749,
+      "step": 13134
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0001614072229140722,
+      "loss": 0.1663,
+      "step": 13135
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0001613947696139477,
+      "loss": 0.1426,
+      "step": 13136
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016138231631382315,
+      "loss": 0.1476,
+      "step": 13137
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016136986301369864,
+      "loss": 0.1669,
+      "step": 13138
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016135740971357407,
+      "loss": 0.259,
+      "step": 13139
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016134495641344956,
+      "loss": 0.1931,
+      "step": 13140
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016133250311332501,
+      "loss": 0.15,
+      "step": 13141
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016132004981320047,
+      "loss": 0.1542,
+      "step": 13142
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016130759651307596,
+      "loss": 0.181,
+      "step": 13143
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0001612951432129514,
+      "loss": 0.1108,
+      "step": 13144
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016128268991282688,
+      "loss": 0.2526,
+      "step": 13145
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016127023661270234,
+      "loss": 0.1657,
+      "step": 13146
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016125778331257782,
+      "loss": 0.2222,
+      "step": 13147
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016124533001245328,
+      "loss": 0.14,
+      "step": 13148
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016123287671232877,
+      "loss": 0.2273,
+      "step": 13149
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0001612204234122042,
+      "loss": 0.2253,
+      "step": 13150
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0001612079701120797,
+      "loss": 0.2183,
+      "step": 13151
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016119551681195515,
+      "loss": 0.4659,
+      "step": 13152
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016118306351183063,
+      "loss": 0.2735,
+      "step": 13153
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0001611706102117061,
+      "loss": 0.1785,
+      "step": 13154
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016115815691158158,
+      "loss": 0.2688,
+      "step": 13155
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.000161145703611457,
+      "loss": 0.2044,
+      "step": 13156
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016113325031133247,
+      "loss": 0.1938,
+      "step": 13157
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016112079701120796,
+      "loss": 0.1976,
+      "step": 13158
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016110834371108342,
+      "loss": 0.1709,
+      "step": 13159
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0001610958904109589,
+      "loss": 0.1401,
+      "step": 13160
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016108343711083433,
+      "loss": 0.0986,
+      "step": 13161
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00016107098381070982,
+      "loss": 0.1134,
+      "step": 13162
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016105853051058528,
+      "loss": 0.1281,
+      "step": 13163
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016104607721046077,
+      "loss": 0.2877,
+      "step": 13164
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016103362391033622,
+      "loss": 0.2216,
+      "step": 13165
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0001610211706102117,
+      "loss": 0.1421,
+      "step": 13166
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016100871731008714,
+      "loss": 0.1519,
+      "step": 13167
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016099626400996263,
+      "loss": 0.1794,
+      "step": 13168
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0001609838107098381,
+      "loss": 0.1678,
+      "step": 13169
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016097135740971357,
+      "loss": 0.2651,
+      "step": 13170
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016095890410958903,
+      "loss": 0.2784,
+      "step": 13171
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016094645080946447,
+      "loss": 0.1582,
+      "step": 13172
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016093399750933995,
+      "loss": 0.1789,
+      "step": 13173
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0001609215442092154,
+      "loss": 0.1402,
+      "step": 13174
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0001609090909090909,
+      "loss": 0.158,
+      "step": 13175
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016089663760896636,
+      "loss": 0.1128,
+      "step": 13176
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016088418430884184,
+      "loss": 0.2732,
+      "step": 13177
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016087173100871727,
+      "loss": 0.2244,
+      "step": 13178
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016085927770859276,
+      "loss": 0.212,
+      "step": 13179
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016084682440846822,
+      "loss": 0.1724,
+      "step": 13180
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0001608343711083437,
+      "loss": 0.1386,
+      "step": 13181
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016082191780821917,
+      "loss": 0.1621,
+      "step": 13182
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016080946450809465,
+      "loss": 0.1931,
+      "step": 13183
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016079701120797008,
+      "loss": 0.225,
+      "step": 13184
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016078455790784557,
+      "loss": 0.1315,
+      "step": 13185
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016077210460772103,
+      "loss": 0.1479,
+      "step": 13186
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016075965130759652,
+      "loss": 0.1576,
+      "step": 13187
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00016074719800747198,
+      "loss": 0.1052,
+      "step": 13188
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0001607347447073474,
+      "loss": 0.4129,
+      "step": 13189
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0001607222914072229,
+      "loss": 0.1905,
+      "step": 13190
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016070983810709835,
+      "loss": 0.2326,
+      "step": 13191
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016069738480697384,
+      "loss": 0.1571,
+      "step": 13192
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0001606849315068493,
+      "loss": 0.1783,
+      "step": 13193
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016067247820672478,
+      "loss": 0.1186,
+      "step": 13194
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016066002490660022,
+      "loss": 0.2394,
+      "step": 13195
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0001606475716064757,
+      "loss": 0.433,
+      "step": 13196
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016063511830635116,
+      "loss": 0.2123,
+      "step": 13197
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016062266500622665,
+      "loss": 0.1646,
+      "step": 13198
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0001606102117061021,
+      "loss": 0.1828,
+      "step": 13199
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0001605977584059776,
+      "loss": 0.119,
+      "step": 13200
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016058530510585303,
+      "loss": 0.2737,
+      "step": 13201
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0001605728518057285,
+      "loss": 0.2626,
+      "step": 13202
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016056039850560397,
+      "loss": 0.2221,
+      "step": 13203
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016054794520547943,
+      "loss": 0.2321,
+      "step": 13204
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016053549190535492,
+      "loss": 0.181,
+      "step": 13205
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016052303860523035,
+      "loss": 0.1206,
+      "step": 13206
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016051058530510583,
+      "loss": 0.203,
+      "step": 13207
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0001604981320049813,
+      "loss": 0.2622,
+      "step": 13208
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016048567870485678,
+      "loss": 0.1775,
+      "step": 13209
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016047322540473224,
+      "loss": 0.1469,
+      "step": 13210
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016046077210460773,
+      "loss": 0.1873,
+      "step": 13211
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016044831880448316,
+      "loss": 0.1625,
+      "step": 13212
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00016043586550435864,
+      "loss": 0.2433,
+      "step": 13213
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0001604234122042341,
+      "loss": 0.244,
+      "step": 13214
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0001604109589041096,
+      "loss": 0.2291,
+      "step": 13215
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016039850560398505,
+      "loss": 0.2017,
+      "step": 13216
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016038605230386053,
+      "loss": 0.295,
+      "step": 13217
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016037359900373597,
+      "loss": 0.1493,
+      "step": 13218
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016036114570361143,
+      "loss": 0.2148,
+      "step": 13219
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0001603486924034869,
+      "loss": 0.2962,
+      "step": 13220
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016033623910336237,
+      "loss": 0.2193,
+      "step": 13221
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016032378580323786,
+      "loss": 0.191,
+      "step": 13222
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0001603113325031133,
+      "loss": 0.2093,
+      "step": 13223
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016029887920298878,
+      "loss": 0.1234,
+      "step": 13224
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016028642590286424,
+      "loss": 0.1285,
+      "step": 13225
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016027397260273972,
+      "loss": 0.3432,
+      "step": 13226
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016026151930261518,
+      "loss": 0.2167,
+      "step": 13227
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016024906600249067,
+      "loss": 0.1394,
+      "step": 13228
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0001602366127023661,
+      "loss": 0.2421,
+      "step": 13229
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016022415940224159,
+      "loss": 0.2609,
+      "step": 13230
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016021170610211704,
+      "loss": 0.1391,
+      "step": 13231
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016019925280199253,
+      "loss": 0.1506,
+      "step": 13232
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.000160186799501868,
+      "loss": 0.4682,
+      "step": 13233
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016017434620174348,
+      "loss": 0.2007,
+      "step": 13234
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0001601618929016189,
+      "loss": 0.1452,
+      "step": 13235
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016014943960149437,
+      "loss": 0.1842,
+      "step": 13236
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016013698630136985,
+      "loss": 0.1147,
+      "step": 13237
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0001601245330012453,
+      "loss": 0.1303,
+      "step": 13238
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0001601120797011208,
+      "loss": 0.413,
+      "step": 13239
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00016009962640099623,
+      "loss": 0.2169,
+      "step": 13240
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00016008717310087172,
+      "loss": 0.1638,
+      "step": 13241
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00016007471980074718,
+      "loss": 0.1853,
+      "step": 13242
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00016006226650062266,
+      "loss": 0.1112,
+      "step": 13243
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00016004981320049812,
+      "loss": 0.1232,
+      "step": 13244
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0001600373599003736,
+      "loss": 0.2802,
+      "step": 13245
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00016002490660024904,
+      "loss": 0.2014,
+      "step": 13246
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00016001245330012453,
+      "loss": 0.1746,
+      "step": 13247
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015999999999999999,
+      "loss": 0.1579,
+      "step": 13248
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015998754669987547,
+      "loss": 0.1738,
+      "step": 13249
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015997509339975093,
+      "loss": 0.1278,
+      "step": 13250
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015996264009962636,
+      "loss": 0.1068,
+      "step": 13251
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015995018679950185,
+      "loss": 0.1847,
+      "step": 13252
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0001599377334993773,
+      "loss": 0.1399,
+      "step": 13253
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0001599252801992528,
+      "loss": 0.1821,
+      "step": 13254
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015991282689912825,
+      "loss": 0.1439,
+      "step": 13255
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015990037359900374,
+      "loss": 0.1186,
+      "step": 13256
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015988792029887917,
+      "loss": 0.2165,
+      "step": 13257
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015987546699875466,
+      "loss": 0.4425,
+      "step": 13258
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015986301369863012,
+      "loss": 0.1991,
+      "step": 13259
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0001598505603985056,
+      "loss": 0.1225,
+      "step": 13260
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015983810709838106,
+      "loss": 0.1681,
+      "step": 13261
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015982565379825655,
+      "loss": 0.2112,
+      "step": 13262
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015981320049813198,
+      "loss": 0.1558,
+      "step": 13263
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015980074719800747,
+      "loss": 0.2535,
+      "step": 13264
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.00015978829389788293,
+      "loss": 0.3016,
+      "step": 13265
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0001597758405977584,
+      "loss": 0.2458,
+      "step": 13266
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015976338729763387,
+      "loss": 0.1465,
+      "step": 13267
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0001597509339975093,
+      "loss": 0.152,
+      "step": 13268
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0001597384806973848,
+      "loss": 0.153,
+      "step": 13269
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015972602739726025,
+      "loss": 0.2093,
+      "step": 13270
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015971357409713574,
+      "loss": 0.2379,
+      "step": 13271
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0001597011207970112,
+      "loss": 0.2007,
+      "step": 13272
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015968866749688668,
+      "loss": 0.1404,
+      "step": 13273
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015967621419676211,
+      "loss": 0.1746,
+      "step": 13274
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0001596637608966376,
+      "loss": 0.3238,
+      "step": 13275
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015965130759651306,
+      "loss": 0.2688,
+      "step": 13276
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015963885429638855,
+      "loss": 0.3833,
+      "step": 13277
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.000159626400996264,
+      "loss": 0.3488,
+      "step": 13278
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015961394769613946,
+      "loss": 0.2161,
+      "step": 13279
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015960149439601492,
+      "loss": 0.1721,
+      "step": 13280
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0001595890410958904,
+      "loss": 0.2689,
+      "step": 13281
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015957658779576587,
+      "loss": 0.1817,
+      "step": 13282
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015956413449564133,
+      "loss": 0.2717,
+      "step": 13283
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0001595516811955168,
+      "loss": 0.2442,
+      "step": 13284
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015953922789539225,
+      "loss": 0.2133,
+      "step": 13285
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015952677459526773,
+      "loss": 0.1902,
+      "step": 13286
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0001595143212951432,
+      "loss": 0.167,
+      "step": 13287
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015950186799501868,
+      "loss": 0.1972,
+      "step": 13288
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015948941469489414,
+      "loss": 0.241,
+      "step": 13289
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0001594769613947696,
+      "loss": 0.174,
+      "step": 13290
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015946450809464506,
+      "loss": 0.1693,
+      "step": 13291
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00015945205479452054,
+      "loss": 0.1131,
+      "step": 13292
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.000159439601494396,
+      "loss": 0.1161,
+      "step": 13293
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0001594271481942715,
+      "loss": 0.1422,
+      "step": 13294
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015941469489414692,
+      "loss": 0.2292,
+      "step": 13295
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0001594022415940224,
+      "loss": 0.2252,
+      "step": 13296
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015938978829389786,
+      "loss": 0.3297,
+      "step": 13297
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015937733499377332,
+      "loss": 0.1253,
+      "step": 13298
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0001593648816936488,
+      "loss": 0.2495,
+      "step": 13299
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015935242839352427,
+      "loss": 0.216,
+      "step": 13300
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015933997509339973,
+      "loss": 0.2008,
+      "step": 13301
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0001593275217932752,
+      "loss": 0.3454,
+      "step": 13302
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015931506849315067,
+      "loss": 0.288,
+      "step": 13303
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015930261519302613,
+      "loss": 0.2172,
+      "step": 13304
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015929016189290162,
+      "loss": 0.142,
+      "step": 13305
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015927770859277705,
+      "loss": 0.1768,
+      "step": 13306
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015926525529265254,
+      "loss": 0.1704,
+      "step": 13307
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.000159252801992528,
+      "loss": 0.2568,
+      "step": 13308
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015924034869240348,
+      "loss": 0.1867,
+      "step": 13309
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015922789539227894,
+      "loss": 0.2062,
+      "step": 13310
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015921544209215443,
+      "loss": 0.1067,
+      "step": 13311
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015920298879202986,
+      "loss": 0.1651,
+      "step": 13312
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015919053549190532,
+      "loss": 0.2055,
+      "step": 13313
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0001591780821917808,
+      "loss": 0.25,
+      "step": 13314
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015916562889165627,
+      "loss": 0.1888,
+      "step": 13315
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015915317559153175,
+      "loss": 0.1845,
+      "step": 13316
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015914072229140718,
+      "loss": 0.1275,
+      "step": 13317
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00015912826899128267,
+      "loss": 0.1541,
+      "step": 13318
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015911581569115813,
+      "loss": 0.1345,
+      "step": 13319
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015910336239103362,
+      "loss": 0.1752,
+      "step": 13320
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015909090909090907,
+      "loss": 0.1742,
+      "step": 13321
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015907845579078456,
+      "loss": 0.2178,
+      "step": 13322
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015906600249066,
+      "loss": 0.2771,
+      "step": 13323
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015905354919053548,
+      "loss": 0.2055,
+      "step": 13324
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015904109589041094,
+      "loss": 0.2219,
+      "step": 13325
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015902864259028642,
+      "loss": 0.2507,
+      "step": 13326
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015901618929016188,
+      "loss": 0.2987,
+      "step": 13327
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015900373599003737,
+      "loss": 0.2154,
+      "step": 13328
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0001589912826899128,
+      "loss": 0.1272,
+      "step": 13329
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015897882938978826,
+      "loss": 0.103,
+      "step": 13330
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015896637608966375,
+      "loss": 0.1312,
+      "step": 13331
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0001589539227895392,
+      "loss": 0.2194,
+      "step": 13332
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0001589414694894147,
+      "loss": 0.3595,
+      "step": 13333
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015892901618929013,
+      "loss": 0.2456,
+      "step": 13334
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0001589165628891656,
+      "loss": 0.2439,
+      "step": 13335
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015890410958904107,
+      "loss": 0.2193,
+      "step": 13336
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015889165628891656,
+      "loss": 0.127,
+      "step": 13337
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015887920298879202,
+      "loss": 0.1058,
+      "step": 13338
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0001588667496886675,
+      "loss": 0.3459,
+      "step": 13339
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015885429638854293,
+      "loss": 0.2017,
+      "step": 13340
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015884184308841842,
+      "loss": 0.1547,
+      "step": 13341
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015882938978829388,
+      "loss": 0.1366,
+      "step": 13342
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015881693648816937,
+      "loss": 0.2183,
+      "step": 13343
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015880448318804483,
+      "loss": 0.1453,
+      "step": 13344
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00015879202988792026,
+      "loss": 0.3277,
+      "step": 13345
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015877957658779574,
+      "loss": 0.2389,
+      "step": 13346
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0001587671232876712,
+      "loss": 0.1954,
+      "step": 13347
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0001587546699875467,
+      "loss": 0.1885,
+      "step": 13348
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015874221668742215,
+      "loss": 0.1392,
+      "step": 13349
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015872976338729763,
+      "loss": 0.155,
+      "step": 13350
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015871731008717307,
+      "loss": 0.2261,
+      "step": 13351
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015870485678704855,
+      "loss": 0.2238,
+      "step": 13352
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.000158692403486924,
+      "loss": 0.2805,
+      "step": 13353
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0001586799501867995,
+      "loss": 0.1504,
+      "step": 13354
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015866749688667496,
+      "loss": 0.1422,
+      "step": 13355
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015865504358655044,
+      "loss": 0.114,
+      "step": 13356
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015864259028642588,
+      "loss": 0.1432,
+      "step": 13357
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015863013698630136,
+      "loss": 0.3871,
+      "step": 13358
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015861768368617682,
+      "loss": 0.1463,
+      "step": 13359
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0001586052303860523,
+      "loss": 0.1837,
+      "step": 13360
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015859277708592777,
+      "loss": 0.237,
+      "step": 13361
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0001585803237858032,
+      "loss": 0.1917,
+      "step": 13362
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015856787048567868,
+      "loss": 0.2849,
+      "step": 13363
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015855541718555414,
+      "loss": 0.2495,
+      "step": 13364
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015854296388542963,
+      "loss": 0.2406,
+      "step": 13365
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0001585305105853051,
+      "loss": 0.16,
+      "step": 13366
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015851805728518058,
+      "loss": 0.2146,
+      "step": 13367
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.000158505603985056,
+      "loss": 0.2307,
+      "step": 13368
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0001584931506849315,
+      "loss": 0.1647,
+      "step": 13369
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015848069738480695,
+      "loss": 0.2866,
+      "step": 13370
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.00015846824408468244,
+      "loss": 0.2307,
+      "step": 13371
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0001584557907845579,
+      "loss": 0.185,
+      "step": 13372
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015844333748443338,
+      "loss": 0.1567,
+      "step": 13373
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015843088418430882,
+      "loss": 0.1604,
+      "step": 13374
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0001584184308841843,
+      "loss": 0.1644,
+      "step": 13375
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015840597758405976,
+      "loss": 0.2257,
+      "step": 13376
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015839352428393522,
+      "loss": 0.2826,
+      "step": 13377
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0001583810709838107,
+      "loss": 0.2337,
+      "step": 13378
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015836861768368614,
+      "loss": 0.2605,
+      "step": 13379
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015835616438356163,
+      "loss": 0.129,
+      "step": 13380
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015834371108343709,
+      "loss": 0.1476,
+      "step": 13381
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015833125778331257,
+      "loss": 0.1966,
+      "step": 13382
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015831880448318803,
+      "loss": 0.2327,
+      "step": 13383
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015830635118306352,
+      "loss": 0.2008,
+      "step": 13384
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015829389788293895,
+      "loss": 0.152,
+      "step": 13385
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015828144458281444,
+      "loss": 0.1363,
+      "step": 13386
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0001582689912826899,
+      "loss": 0.2289,
+      "step": 13387
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015825653798256538,
+      "loss": 0.1673,
+      "step": 13388
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015824408468244084,
+      "loss": 0.3062,
+      "step": 13389
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015823163138231633,
+      "loss": 0.1852,
+      "step": 13390
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015821917808219176,
+      "loss": 0.1761,
+      "step": 13391
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015820672478206722,
+      "loss": 0.1802,
+      "step": 13392
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0001581942714819427,
+      "loss": 0.1419,
+      "step": 13393
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015818181818181816,
+      "loss": 0.1717,
+      "step": 13394
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015816936488169365,
+      "loss": 0.2069,
+      "step": 13395
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015815691158156908,
+      "loss": 0.2447,
+      "step": 13396
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00015814445828144457,
+      "loss": 0.223,
+      "step": 13397
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015813200498132003,
+      "loss": 0.1708,
+      "step": 13398
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0001581195516811955,
+      "loss": 0.1654,
+      "step": 13399
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015810709838107097,
+      "loss": 0.1468,
+      "step": 13400
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015809464508094646,
+      "loss": 0.166,
+      "step": 13401
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0001580821917808219,
+      "loss": 0.3079,
+      "step": 13402
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015806973848069738,
+      "loss": 0.2495,
+      "step": 13403
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015805728518057284,
+      "loss": 0.2175,
+      "step": 13404
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015804483188044832,
+      "loss": 0.2364,
+      "step": 13405
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015803237858032378,
+      "loss": 0.2169,
+      "step": 13406
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015801992528019927,
+      "loss": 0.3122,
+      "step": 13407
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0001580074719800747,
+      "loss": 0.3113,
+      "step": 13408
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015799501867995016,
+      "loss": 0.171,
+      "step": 13409
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015798256537982565,
+      "loss": 0.1756,
+      "step": 13410
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0001579701120797011,
+      "loss": 0.1835,
+      "step": 13411
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0001579576587795766,
+      "loss": 0.1701,
+      "step": 13412
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015794520547945202,
+      "loss": 0.2521,
+      "step": 13413
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0001579327521793275,
+      "loss": 0.2773,
+      "step": 13414
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015792029887920297,
+      "loss": 0.1997,
+      "step": 13415
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015790784557907845,
+      "loss": 0.2145,
+      "step": 13416
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015789539227895391,
+      "loss": 0.2591,
+      "step": 13417
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0001578829389788294,
+      "loss": 0.154,
+      "step": 13418
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015787048567870483,
+      "loss": 0.2795,
+      "step": 13419
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015785803237858032,
+      "loss": 0.4739,
+      "step": 13420
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015784557907845578,
+      "loss": 0.2402,
+      "step": 13421
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015783312577833126,
+      "loss": 0.2246,
+      "step": 13422
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00015782067247820672,
+      "loss": 0.1833,
+      "step": 13423
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015780821917808215,
+      "loss": 0.1782,
+      "step": 13424
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015779576587795764,
+      "loss": 0.1403,
+      "step": 13425
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0001577833125778331,
+      "loss": 0.1694,
+      "step": 13426
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0001577708592777086,
+      "loss": 0.3297,
+      "step": 13427
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015775840597758405,
+      "loss": 0.1679,
+      "step": 13428
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015774595267745953,
+      "loss": 0.192,
+      "step": 13429
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015773349937733496,
+      "loss": 0.123,
+      "step": 13430
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015772104607721045,
+      "loss": 0.1766,
+      "step": 13431
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0001577085927770859,
+      "loss": 0.1496,
+      "step": 13432
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0001576961394769614,
+      "loss": 0.2675,
+      "step": 13433
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015768368617683686,
+      "loss": 0.2635,
+      "step": 13434
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015767123287671234,
+      "loss": 0.1583,
+      "step": 13435
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015765877957658777,
+      "loss": 0.3303,
+      "step": 13436
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015764632627646326,
+      "loss": 0.1505,
+      "step": 13437
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015763387297633872,
+      "loss": 0.2123,
+      "step": 13438
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015762141967621418,
+      "loss": 0.2685,
+      "step": 13439
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015760896637608966,
+      "loss": 0.1686,
+      "step": 13440
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0001575965130759651,
+      "loss": 0.1889,
+      "step": 13441
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015758405977584058,
+      "loss": 0.2398,
+      "step": 13442
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015757160647571604,
+      "loss": 0.1753,
+      "step": 13443
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015755915317559153,
+      "loss": 0.1149,
+      "step": 13444
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.000157546699875467,
+      "loss": 0.2551,
+      "step": 13445
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015753424657534245,
+      "loss": 0.2755,
+      "step": 13446
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0001575217932752179,
+      "loss": 0.1785,
+      "step": 13447
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0001575093399750934,
+      "loss": 0.1493,
+      "step": 13448
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00015749688667496885,
+      "loss": 0.1222,
+      "step": 13449
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015748443337484434,
+      "loss": 0.2623,
+      "step": 13450
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0001574719800747198,
+      "loss": 0.1943,
+      "step": 13451
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015745952677459526,
+      "loss": 0.2383,
+      "step": 13452
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015744707347447071,
+      "loss": 0.201,
+      "step": 13453
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0001574346201743462,
+      "loss": 0.1629,
+      "step": 13454
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015742216687422166,
+      "loss": 0.2123,
+      "step": 13455
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015740971357409712,
+      "loss": 0.1752,
+      "step": 13456
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015739726027397258,
+      "loss": 0.3,
+      "step": 13457
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015738480697384804,
+      "loss": 0.2944,
+      "step": 13458
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015737235367372352,
+      "loss": 0.253,
+      "step": 13459
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015735990037359898,
+      "loss": 0.1626,
+      "step": 13460
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015734744707347447,
+      "loss": 0.1619,
+      "step": 13461
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015733499377334993,
+      "loss": 0.2925,
+      "step": 13462
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0001573225404732254,
+      "loss": 0.1396,
+      "step": 13463
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015731008717310085,
+      "loss": 0.3121,
+      "step": 13464
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015729763387297633,
+      "loss": 0.1624,
+      "step": 13465
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0001572851805728518,
+      "loss": 0.1826,
+      "step": 13466
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015727272727272728,
+      "loss": 0.1738,
+      "step": 13467
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0001572602739726027,
+      "loss": 0.1179,
+      "step": 13468
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0001572478206724782,
+      "loss": 0.1691,
+      "step": 13469
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015723536737235366,
+      "loss": 0.283,
+      "step": 13470
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015722291407222912,
+      "loss": 0.2172,
+      "step": 13471
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0001572104607721046,
+      "loss": 0.218,
+      "step": 13472
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015719800747198006,
+      "loss": 0.1362,
+      "step": 13473
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015718555417185552,
+      "loss": 0.2512,
+      "step": 13474
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00015717310087173098,
+      "loss": 0.115,
+      "step": 13475
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015716064757160647,
+      "loss": 0.1852,
+      "step": 13476
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015714819427148192,
+      "loss": 0.1641,
+      "step": 13477
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0001571357409713574,
+      "loss": 0.2286,
+      "step": 13478
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015712328767123284,
+      "loss": 0.2801,
+      "step": 13479
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015711083437110833,
+      "loss": 0.2747,
+      "step": 13480
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0001570983810709838,
+      "loss": 0.1688,
+      "step": 13481
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015708592777085927,
+      "loss": 0.2528,
+      "step": 13482
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015707347447073473,
+      "loss": 0.251,
+      "step": 13483
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015706102117061022,
+      "loss": 0.1879,
+      "step": 13484
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015704856787048565,
+      "loss": 0.2707,
+      "step": 13485
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0001570361145703611,
+      "loss": 0.1983,
+      "step": 13486
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0001570236612702366,
+      "loss": 0.1959,
+      "step": 13487
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015701120797011206,
+      "loss": 0.1676,
+      "step": 13488
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015699875466998754,
+      "loss": 0.3299,
+      "step": 13489
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015698630136986298,
+      "loss": 0.2426,
+      "step": 13490
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015697384806973846,
+      "loss": 0.2199,
+      "step": 13491
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015696139476961392,
+      "loss": 0.1601,
+      "step": 13492
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0001569489414694894,
+      "loss": 0.1909,
+      "step": 13493
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015693648816936487,
+      "loss": 0.226,
+      "step": 13494
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015692403486924035,
+      "loss": 0.2179,
+      "step": 13495
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015691158156911578,
+      "loss": 0.2219,
+      "step": 13496
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015689912826899127,
+      "loss": 0.1442,
+      "step": 13497
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015688667496886673,
+      "loss": 0.1356,
+      "step": 13498
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015687422166874222,
+      "loss": 0.1623,
+      "step": 13499
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015686176836861768,
+      "loss": 0.1552,
+      "step": 13500
+    },
+    {
+      "epoch": 5.17,
+      "eval_cer": 0.10586651579296487,
+      "eval_loss": 0.3719942569732666,
+      "eval_runtime": 47.2624,
+      "eval_samples_per_second": 37.091,
+      "eval_steps_per_second": 4.655,
+      "eval_wer": 0.17289677331764494,
+      "step": 13500
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00015684931506849316,
+      "loss": 0.1734,
+      "step": 13501
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0001568368617683686,
+      "loss": 0.2012,
+      "step": 13502
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015682440846824405,
+      "loss": 0.2043,
+      "step": 13503
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015681195516811954,
+      "loss": 0.1929,
+      "step": 13504
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.000156799501867995,
+      "loss": 0.1234,
+      "step": 13505
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015678704856787048,
+      "loss": 0.1505,
+      "step": 13506
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015677459526774592,
+      "loss": 0.3018,
+      "step": 13507
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0001567621419676214,
+      "loss": 0.2301,
+      "step": 13508
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015674968866749686,
+      "loss": 0.1976,
+      "step": 13509
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015673723536737235,
+      "loss": 0.3421,
+      "step": 13510
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0001567247820672478,
+      "loss": 0.1885,
+      "step": 13511
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0001567123287671233,
+      "loss": 0.195,
+      "step": 13512
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015669987546699873,
+      "loss": 0.1641,
+      "step": 13513
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0001566874221668742,
+      "loss": 0.2183,
+      "step": 13514
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015667496886674967,
+      "loss": 0.1868,
+      "step": 13515
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015666251556662516,
+      "loss": 0.2881,
+      "step": 13516
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015665006226650062,
+      "loss": 0.164,
+      "step": 13517
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015663760896637605,
+      "loss": 0.2237,
+      "step": 13518
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015662515566625153,
+      "loss": 0.1549,
+      "step": 13519
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.000156612702366127,
+      "loss": 0.3911,
+      "step": 13520
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015660024906600248,
+      "loss": 0.3352,
+      "step": 13521
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015658779576587794,
+      "loss": 0.1685,
+      "step": 13522
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015657534246575343,
+      "loss": 0.1902,
+      "step": 13523
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015656288916562886,
+      "loss": 0.2947,
+      "step": 13524
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00015655043586550434,
+      "loss": 0.1441,
+      "step": 13525
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0001565379825653798,
+      "loss": 0.1586,
+      "step": 13526
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0001565255292652553,
+      "loss": 0.2751,
+      "step": 13527
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015651307596513075,
+      "loss": 0.2271,
+      "step": 13528
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015650062266500624,
+      "loss": 0.1209,
+      "step": 13529
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015648816936488167,
+      "loss": 0.2193,
+      "step": 13530
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015647571606475715,
+      "loss": 0.2117,
+      "step": 13531
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0001564632627646326,
+      "loss": 0.2378,
+      "step": 13532
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015645080946450807,
+      "loss": 0.2786,
+      "step": 13533
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015643835616438356,
+      "loss": 0.2192,
+      "step": 13534
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.000156425902864259,
+      "loss": 0.3144,
+      "step": 13535
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015641344956413448,
+      "loss": 0.2054,
+      "step": 13536
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015640099626400994,
+      "loss": 0.133,
+      "step": 13537
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015638854296388542,
+      "loss": 0.1561,
+      "step": 13538
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015637608966376088,
+      "loss": 0.2432,
+      "step": 13539
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015636363636363637,
+      "loss": 0.238,
+      "step": 13540
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0001563511830635118,
+      "loss": 0.1412,
+      "step": 13541
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015633872976338729,
+      "loss": 0.1828,
+      "step": 13542
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015632627646326274,
+      "loss": 0.2023,
+      "step": 13543
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015631382316313823,
+      "loss": 0.1267,
+      "step": 13544
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0001563013698630137,
+      "loss": 0.3042,
+      "step": 13545
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015628891656288918,
+      "loss": 0.1978,
+      "step": 13546
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0001562764632627646,
+      "loss": 0.1885,
+      "step": 13547
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0001562640099626401,
+      "loss": 0.1965,
+      "step": 13548
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015625155666251555,
+      "loss": 0.1532,
+      "step": 13549
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.000156239103362391,
+      "loss": 0.1825,
+      "step": 13550
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0001562266500622665,
+      "loss": 0.2773,
+      "step": 13551
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015621419676214193,
+      "loss": 0.2722,
+      "step": 13552
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00015620174346201742,
+      "loss": 0.144,
+      "step": 13553
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015618929016189288,
+      "loss": 0.1587,
+      "step": 13554
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015617683686176836,
+      "loss": 0.1603,
+      "step": 13555
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015616438356164382,
+      "loss": 0.159,
+      "step": 13556
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0001561519302615193,
+      "loss": 0.1347,
+      "step": 13557
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015613947696139474,
+      "loss": 0.2141,
+      "step": 13558
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015612702366127023,
+      "loss": 0.1956,
+      "step": 13559
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015611457036114569,
+      "loss": 0.1678,
+      "step": 13560
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015610211706102117,
+      "loss": 0.1945,
+      "step": 13561
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015608966376089663,
+      "loss": 0.2325,
+      "step": 13562
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015607721046077212,
+      "loss": 0.2638,
+      "step": 13563
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015606475716064755,
+      "loss": 0.4218,
+      "step": 13564
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.000156052303860523,
+      "loss": 0.2263,
+      "step": 13565
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0001560398505603985,
+      "loss": 0.2656,
+      "step": 13566
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015602739726027395,
+      "loss": 0.2432,
+      "step": 13567
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015601494396014944,
+      "loss": 0.1541,
+      "step": 13568
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015600249066002487,
+      "loss": 0.2164,
+      "step": 13569
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015599003735990036,
+      "loss": 0.2182,
+      "step": 13570
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015597758405977582,
+      "loss": 0.1849,
+      "step": 13571
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0001559651307596513,
+      "loss": 0.2328,
+      "step": 13572
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015595267745952676,
+      "loss": 0.1764,
+      "step": 13573
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015594022415940225,
+      "loss": 0.1705,
+      "step": 13574
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015592777085927768,
+      "loss": 0.1422,
+      "step": 13575
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015591531755915317,
+      "loss": 0.2037,
+      "step": 13576
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015590286425902863,
+      "loss": 0.2282,
+      "step": 13577
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015589041095890411,
+      "loss": 0.2412,
+      "step": 13578
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.00015587795765877957,
+      "loss": 0.1601,
+      "step": 13579
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.000155865504358655,
+      "loss": 0.1356,
+      "step": 13580
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0001558530510585305,
+      "loss": 0.1944,
+      "step": 13581
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015584059775840595,
+      "loss": 0.1634,
+      "step": 13582
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015582814445828144,
+      "loss": 0.2696,
+      "step": 13583
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0001558156911581569,
+      "loss": 0.3556,
+      "step": 13584
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015580323785803238,
+      "loss": 0.2054,
+      "step": 13585
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015579078455790781,
+      "loss": 0.152,
+      "step": 13586
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0001557783312577833,
+      "loss": 0.1226,
+      "step": 13587
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015576587795765876,
+      "loss": 0.3198,
+      "step": 13588
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015575342465753425,
+      "loss": 0.3621,
+      "step": 13589
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0001557409713574097,
+      "loss": 0.3012,
+      "step": 13590
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0001557285180572852,
+      "loss": 0.1632,
+      "step": 13591
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015571606475716062,
+      "loss": 0.1493,
+      "step": 13592
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0001557036114570361,
+      "loss": 0.2175,
+      "step": 13593
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015569115815691157,
+      "loss": 0.2,
+      "step": 13594
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015567870485678706,
+      "loss": 0.2915,
+      "step": 13595
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015566625155666251,
+      "loss": 0.3146,
+      "step": 13596
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015565379825653795,
+      "loss": 0.1518,
+      "step": 13597
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015564134495641343,
+      "loss": 0.2078,
+      "step": 13598
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0001556288916562889,
+      "loss": 0.2312,
+      "step": 13599
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015561643835616438,
+      "loss": 0.1537,
+      "step": 13600
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015560398505603984,
+      "loss": 0.2087,
+      "step": 13601
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015559153175591532,
+      "loss": 0.3519,
+      "step": 13602
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015557907845579076,
+      "loss": 0.1815,
+      "step": 13603
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00015556662515566624,
+      "loss": 0.1526,
+      "step": 13604
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0001555541718555417,
+      "loss": 0.158,
+      "step": 13605
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0001555417185554172,
+      "loss": 0.1769,
+      "step": 13606
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015552926525529265,
+      "loss": 0.1747,
+      "step": 13607
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0001555168119551681,
+      "loss": 0.3352,
+      "step": 13608
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015550435865504356,
+      "loss": 0.1961,
+      "step": 13609
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015549190535491905,
+      "loss": 0.1688,
+      "step": 13610
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0001554794520547945,
+      "loss": 0.1503,
+      "step": 13611
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015546699875466997,
+      "loss": 0.1775,
+      "step": 13612
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015545454545454546,
+      "loss": 0.1681,
+      "step": 13613
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0001554420921544209,
+      "loss": 0.3608,
+      "step": 13614
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015542963885429637,
+      "loss": 0.1686,
+      "step": 13615
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015541718555417183,
+      "loss": 0.2076,
+      "step": 13616
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015540473225404732,
+      "loss": 0.2534,
+      "step": 13617
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015539227895392278,
+      "loss": 0.1186,
+      "step": 13618
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015537982565379824,
+      "loss": 0.2002,
+      "step": 13619
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0001553673723536737,
+      "loss": 0.2407,
+      "step": 13620
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015535491905354918,
+      "loss": 0.2613,
+      "step": 13621
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015534246575342464,
+      "loss": 0.1464,
+      "step": 13622
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015533001245330013,
+      "loss": 0.2101,
+      "step": 13623
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0001553175591531756,
+      "loss": 0.1541,
+      "step": 13624
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015530510585305105,
+      "loss": 0.2989,
+      "step": 13625
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0001552926525529265,
+      "loss": 0.1775,
+      "step": 13626
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015528019925280197,
+      "loss": 0.308,
+      "step": 13627
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015526774595267745,
+      "loss": 0.234,
+      "step": 13628
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0001552552926525529,
+      "loss": 0.21,
+      "step": 13629
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015524283935242837,
+      "loss": 0.1526,
+      "step": 13630
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015523038605230383,
+      "loss": 0.1855,
+      "step": 13631
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00015521793275217932,
+      "loss": 0.1471,
+      "step": 13632
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015520547945205477,
+      "loss": 0.2814,
+      "step": 13633
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015519302615193026,
+      "loss": 0.2092,
+      "step": 13634
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015518057285180572,
+      "loss": 0.2516,
+      "step": 13635
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015516811955168118,
+      "loss": 0.1431,
+      "step": 13636
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015515566625155664,
+      "loss": 0.2373,
+      "step": 13637
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015514321295143212,
+      "loss": 0.2159,
+      "step": 13638
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015513075965130758,
+      "loss": 0.4468,
+      "step": 13639
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015511830635118307,
+      "loss": 0.2336,
+      "step": 13640
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0001551058530510585,
+      "loss": 0.1839,
+      "step": 13641
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.000155093399750934,
+      "loss": 0.1505,
+      "step": 13642
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015508094645080945,
+      "loss": 0.1298,
+      "step": 13643
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0001550684931506849,
+      "loss": 0.1421,
+      "step": 13644
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0001550560398505604,
+      "loss": 0.3731,
+      "step": 13645
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015504358655043585,
+      "loss": 0.248,
+      "step": 13646
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0001550311332503113,
+      "loss": 0.1528,
+      "step": 13647
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015501867995018677,
+      "loss": 0.1805,
+      "step": 13648
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015500622665006226,
+      "loss": 0.2338,
+      "step": 13649
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015499377334993772,
+      "loss": 0.1898,
+      "step": 13650
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0001549813200498132,
+      "loss": 0.1939,
+      "step": 13651
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015496886674968863,
+      "loss": 0.3005,
+      "step": 13652
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015495641344956412,
+      "loss": 0.2253,
+      "step": 13653
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015494396014943958,
+      "loss": 0.2644,
+      "step": 13654
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015493150684931507,
+      "loss": 0.1499,
+      "step": 13655
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015491905354919053,
+      "loss": 0.1541,
+      "step": 13656
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.000154906600249066,
+      "loss": 0.2217,
+      "step": 13657
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00015489414694894144,
+      "loss": 0.21,
+      "step": 13658
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0001548816936488169,
+      "loss": 0.1581,
+      "step": 13659
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0001548692403486924,
+      "loss": 0.221,
+      "step": 13660
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015485678704856785,
+      "loss": 0.1271,
+      "step": 13661
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015484433374844333,
+      "loss": 0.1425,
+      "step": 13662
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015483188044831877,
+      "loss": 0.21,
+      "step": 13663
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015481942714819425,
+      "loss": 0.2912,
+      "step": 13664
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0001548069738480697,
+      "loss": 0.1745,
+      "step": 13665
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0001547945205479452,
+      "loss": 0.1851,
+      "step": 13666
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015478206724782066,
+      "loss": 0.1654,
+      "step": 13667
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015476961394769614,
+      "loss": 0.1969,
+      "step": 13668
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015475716064757158,
+      "loss": 0.1419,
+      "step": 13669
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015474470734744706,
+      "loss": 0.1503,
+      "step": 13670
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015473225404732252,
+      "loss": 0.2417,
+      "step": 13671
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.000154719800747198,
+      "loss": 0.3583,
+      "step": 13672
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015470734744707347,
+      "loss": 0.1411,
+      "step": 13673
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015469489414694895,
+      "loss": 0.1573,
+      "step": 13674
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015468244084682438,
+      "loss": 0.2536,
+      "step": 13675
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015466998754669984,
+      "loss": 0.1985,
+      "step": 13676
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015465753424657533,
+      "loss": 0.2022,
+      "step": 13677
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0001546450809464508,
+      "loss": 0.204,
+      "step": 13678
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015463262764632628,
+      "loss": 0.2816,
+      "step": 13679
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0001546201743462017,
+      "loss": 0.1485,
+      "step": 13680
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0001546077210460772,
+      "loss": 0.2446,
+      "step": 13681
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015459526774595265,
+      "loss": 0.1335,
+      "step": 13682
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.00015458281444582814,
+      "loss": 0.3836,
+      "step": 13683
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0001545703611457036,
+      "loss": 0.209,
+      "step": 13684
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015455790784557909,
+      "loss": 0.2593,
+      "step": 13685
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015454545454545452,
+      "loss": 0.212,
+      "step": 13686
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015453300124533,
+      "loss": 0.2,
+      "step": 13687
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015452054794520546,
+      "loss": 0.1653,
+      "step": 13688
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015450809464508095,
+      "loss": 0.2901,
+      "step": 13689
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0001544956413449564,
+      "loss": 0.2337,
+      "step": 13690
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015448318804483184,
+      "loss": 0.2006,
+      "step": 13691
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015447073474470733,
+      "loss": 0.1206,
+      "step": 13692
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015445828144458279,
+      "loss": 0.305,
+      "step": 13693
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015444582814445827,
+      "loss": 0.2044,
+      "step": 13694
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015443337484433373,
+      "loss": 0.2166,
+      "step": 13695
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015442092154420922,
+      "loss": 0.1564,
+      "step": 13696
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015440846824408465,
+      "loss": 0.1465,
+      "step": 13697
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015439601494396014,
+      "loss": 0.1846,
+      "step": 13698
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0001543835616438356,
+      "loss": 0.1552,
+      "step": 13699
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015437110834371108,
+      "loss": 0.1238,
+      "step": 13700
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015435865504358654,
+      "loss": 0.1384,
+      "step": 13701
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015434620174346203,
+      "loss": 0.3936,
+      "step": 13702
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015433374844333746,
+      "loss": 0.2186,
+      "step": 13703
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015432129514321294,
+      "loss": 0.1417,
+      "step": 13704
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0001543088418430884,
+      "loss": 0.1652,
+      "step": 13705
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015429638854296386,
+      "loss": 0.1533,
+      "step": 13706
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015428393524283935,
+      "loss": 0.286,
+      "step": 13707
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015427148194271478,
+      "loss": 0.2617,
+      "step": 13708
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015425902864259027,
+      "loss": 0.2241,
+      "step": 13709
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00015424657534246573,
+      "loss": 0.1507,
+      "step": 13710
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0001542341220423412,
+      "loss": 0.2072,
+      "step": 13711
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015422166874221667,
+      "loss": 0.2835,
+      "step": 13712
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015420921544209216,
+      "loss": 0.1509,
+      "step": 13713
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0001541967621419676,
+      "loss": 0.4244,
+      "step": 13714
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015418430884184308,
+      "loss": 0.1933,
+      "step": 13715
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015417185554171854,
+      "loss": 0.1486,
+      "step": 13716
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015415940224159402,
+      "loss": 0.2359,
+      "step": 13717
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015414694894146948,
+      "loss": 0.1056,
+      "step": 13718
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015413449564134497,
+      "loss": 0.1234,
+      "step": 13719
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0001541220423412204,
+      "loss": 0.2947,
+      "step": 13720
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015410958904109589,
+      "loss": 0.1738,
+      "step": 13721
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015409713574097135,
+      "loss": 0.1647,
+      "step": 13722
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0001540846824408468,
+      "loss": 0.2539,
+      "step": 13723
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0001540722291407223,
+      "loss": 0.179,
+      "step": 13724
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015405977584059772,
+      "loss": 0.1977,
+      "step": 13725
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0001540473225404732,
+      "loss": 0.1776,
+      "step": 13726
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015403486924034867,
+      "loss": 0.265,
+      "step": 13727
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015402241594022415,
+      "loss": 0.1398,
+      "step": 13728
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015400996264009961,
+      "loss": 0.1639,
+      "step": 13729
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0001539975093399751,
+      "loss": 0.1296,
+      "step": 13730
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015398505603985053,
+      "loss": 0.2191,
+      "step": 13731
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015397260273972602,
+      "loss": 0.1406,
+      "step": 13732
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015396014943960148,
+      "loss": 0.2714,
+      "step": 13733
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015394769613947696,
+      "loss": 0.1563,
+      "step": 13734
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00015393524283935242,
+      "loss": 0.22,
+      "step": 13735
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0001539227895392279,
+      "loss": 0.1436,
+      "step": 13736
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015391033623910334,
+      "loss": 0.2578,
+      "step": 13737
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0001538978829389788,
+      "loss": 0.1395,
+      "step": 13738
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0001538854296388543,
+      "loss": 0.2928,
+      "step": 13739
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015387297633872975,
+      "loss": 0.2239,
+      "step": 13740
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015386052303860523,
+      "loss": 0.2293,
+      "step": 13741
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015384806973848066,
+      "loss": 0.1613,
+      "step": 13742
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015383561643835615,
+      "loss": 0.18,
+      "step": 13743
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0001538231631382316,
+      "loss": 0.2391,
+      "step": 13744
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0001538107098381071,
+      "loss": 0.3363,
+      "step": 13745
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015379825653798256,
+      "loss": 0.2319,
+      "step": 13746
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015378580323785804,
+      "loss": 0.2044,
+      "step": 13747
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015377334993773347,
+      "loss": 0.182,
+      "step": 13748
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015376089663760896,
+      "loss": 0.2237,
+      "step": 13749
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015374844333748442,
+      "loss": 0.1743,
+      "step": 13750
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0001537359900373599,
+      "loss": 0.1466,
+      "step": 13751
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015372353673723536,
+      "loss": 0.2762,
+      "step": 13752
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0001537110834371108,
+      "loss": 0.1865,
+      "step": 13753
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015369863013698628,
+      "loss": 0.1234,
+      "step": 13754
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015368617683686174,
+      "loss": 0.2094,
+      "step": 13755
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015367372353673723,
+      "loss": 0.151,
+      "step": 13756
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0001536612702366127,
+      "loss": 0.1885,
+      "step": 13757
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015364881693648817,
+      "loss": 0.3536,
+      "step": 13758
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0001536363636363636,
+      "loss": 0.183,
+      "step": 13759
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0001536239103362391,
+      "loss": 0.1714,
+      "step": 13760
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015361145703611455,
+      "loss": 0.1667,
+      "step": 13761
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00015359900373599004,
+      "loss": 0.1212,
+      "step": 13762
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0001535865504358655,
+      "loss": 0.2106,
+      "step": 13763
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015357409713574098,
+      "loss": 0.2711,
+      "step": 13764
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015356164383561641,
+      "loss": 0.2121,
+      "step": 13765
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0001535491905354919,
+      "loss": 0.3309,
+      "step": 13766
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015353673723536736,
+      "loss": 0.1959,
+      "step": 13767
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015352428393524285,
+      "loss": 0.1743,
+      "step": 13768
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0001535118306351183,
+      "loss": 0.2689,
+      "step": 13769
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015349937733499374,
+      "loss": 0.2936,
+      "step": 13770
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015348692403486922,
+      "loss": 0.1875,
+      "step": 13771
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015347447073474468,
+      "loss": 0.2516,
+      "step": 13772
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015346201743462017,
+      "loss": 0.2357,
+      "step": 13773
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015344956413449563,
+      "loss": 0.1551,
+      "step": 13774
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015343711083437111,
+      "loss": 0.1799,
+      "step": 13775
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015342465753424655,
+      "loss": 0.2411,
+      "step": 13776
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015341220423412203,
+      "loss": 0.2798,
+      "step": 13777
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0001533997509339975,
+      "loss": 0.2395,
+      "step": 13778
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015338729763387298,
+      "loss": 0.4001,
+      "step": 13779
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015337484433374844,
+      "loss": 0.141,
+      "step": 13780
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0001533623910336239,
+      "loss": 0.1278,
+      "step": 13781
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015334993773349936,
+      "loss": 0.2223,
+      "step": 13782
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015333748443337484,
+      "loss": 0.2441,
+      "step": 13783
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0001533250311332503,
+      "loss": 0.2077,
+      "step": 13784
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015331257783312576,
+      "loss": 0.2323,
+      "step": 13785
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015330012453300125,
+      "loss": 0.149,
+      "step": 13786
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015328767123287668,
+      "loss": 0.2077,
+      "step": 13787
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00015327521793275217,
+      "loss": 0.258,
+      "step": 13788
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015326276463262762,
+      "loss": 0.3241,
+      "step": 13789
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0001532503113325031,
+      "loss": 0.287,
+      "step": 13790
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015323785803237857,
+      "loss": 0.303,
+      "step": 13791
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015322540473225403,
+      "loss": 0.1525,
+      "step": 13792
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0001532129514321295,
+      "loss": 0.2117,
+      "step": 13793
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015320049813200497,
+      "loss": 0.3017,
+      "step": 13794
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015318804483188043,
+      "loss": 0.2065,
+      "step": 13795
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015317559153175592,
+      "loss": 0.2524,
+      "step": 13796
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015316313823163138,
+      "loss": 0.1544,
+      "step": 13797
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015315068493150684,
+      "loss": 0.1559,
+      "step": 13798
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0001531382316313823,
+      "loss": 0.1896,
+      "step": 13799
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015312577833125776,
+      "loss": 0.2288,
+      "step": 13800
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015311332503113324,
+      "loss": 0.1566,
+      "step": 13801
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0001531008717310087,
+      "loss": 0.3147,
+      "step": 13802
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015308841843088416,
+      "loss": 0.1913,
+      "step": 13803
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015307596513075962,
+      "loss": 0.1857,
+      "step": 13804
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0001530635118306351,
+      "loss": 0.1007,
+      "step": 13805
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015305105853051057,
+      "loss": 0.1312,
+      "step": 13806
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015303860523038605,
+      "loss": 0.2523,
+      "step": 13807
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0001530261519302615,
+      "loss": 0.2699,
+      "step": 13808
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015301369863013697,
+      "loss": 0.2535,
+      "step": 13809
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015300124533001243,
+      "loss": 0.1346,
+      "step": 13810
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015298879202988792,
+      "loss": 0.2177,
+      "step": 13811
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015297633872976338,
+      "loss": 0.1979,
+      "step": 13812
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.00015296388542963886,
+      "loss": 0.1976,
+      "step": 13813
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0001529514321295143,
+      "loss": 0.2381,
+      "step": 13814
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015293897882938978,
+      "loss": 0.176,
+      "step": 13815
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015292652552926524,
+      "loss": 0.3197,
+      "step": 13816
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0001529140722291407,
+      "loss": 0.1738,
+      "step": 13817
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015290161892901618,
+      "loss": 0.167,
+      "step": 13818
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015288916562889162,
+      "loss": 0.1235,
+      "step": 13819
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0001528767123287671,
+      "loss": 0.3346,
+      "step": 13820
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015286425902864256,
+      "loss": 0.3584,
+      "step": 13821
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015285180572851805,
+      "loss": 0.1392,
+      "step": 13822
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0001528393524283935,
+      "loss": 0.1618,
+      "step": 13823
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.000152826899128269,
+      "loss": 0.1363,
+      "step": 13824
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015281444582814443,
+      "loss": 0.1746,
+      "step": 13825
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0001528019925280199,
+      "loss": 0.2411,
+      "step": 13826
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015278953922789537,
+      "loss": 0.241,
+      "step": 13827
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015277708592777086,
+      "loss": 0.2423,
+      "step": 13828
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015276463262764632,
+      "loss": 0.1736,
+      "step": 13829
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0001527521793275218,
+      "loss": 0.1864,
+      "step": 13830
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015273972602739724,
+      "loss": 0.2527,
+      "step": 13831
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0001527272727272727,
+      "loss": 0.1539,
+      "step": 13832
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015271481942714818,
+      "loss": 0.3156,
+      "step": 13833
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015270236612702364,
+      "loss": 0.1934,
+      "step": 13834
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015268991282689913,
+      "loss": 0.1655,
+      "step": 13835
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015267745952677456,
+      "loss": 0.1845,
+      "step": 13836
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015266500622665004,
+      "loss": 0.2079,
+      "step": 13837
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0001526525529265255,
+      "loss": 0.1538,
+      "step": 13838
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.000152640099626401,
+      "loss": 0.1343,
+      "step": 13839
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00015262764632627645,
+      "loss": 0.2283,
+      "step": 13840
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015261519302615194,
+      "loss": 0.1667,
+      "step": 13841
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015260273972602737,
+      "loss": 0.1362,
+      "step": 13842
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015259028642590285,
+      "loss": 0.1324,
+      "step": 13843
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0001525778331257783,
+      "loss": 0.2891,
+      "step": 13844
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0001525653798256538,
+      "loss": 0.2461,
+      "step": 13845
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015255292652552926,
+      "loss": 0.2237,
+      "step": 13846
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0001525404732254047,
+      "loss": 0.1828,
+      "step": 13847
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015252801992528018,
+      "loss": 0.1256,
+      "step": 13848
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015251556662515564,
+      "loss": 0.1494,
+      "step": 13849
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015250311332503112,
+      "loss": 0.229,
+      "step": 13850
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015249066002490658,
+      "loss": 0.1535,
+      "step": 13851
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015247820672478207,
+      "loss": 0.2908,
+      "step": 13852
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0001524657534246575,
+      "loss": 0.2144,
+      "step": 13853
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015245330012453299,
+      "loss": 0.1518,
+      "step": 13854
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015244084682440844,
+      "loss": 0.1757,
+      "step": 13855
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015242839352428393,
+      "loss": 0.1423,
+      "step": 13856
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0001524159402241594,
+      "loss": 0.1856,
+      "step": 13857
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015240348692403488,
+      "loss": 0.206,
+      "step": 13858
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0001523910336239103,
+      "loss": 0.1931,
+      "step": 13859
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0001523785803237858,
+      "loss": 0.2057,
+      "step": 13860
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015236612702366125,
+      "loss": 0.1784,
+      "step": 13861
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015235367372353674,
+      "loss": 0.1486,
+      "step": 13862
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0001523412204234122,
+      "loss": 0.2037,
+      "step": 13863
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015232876712328763,
+      "loss": 0.2257,
+      "step": 13864
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015231631382316312,
+      "loss": 0.2182,
+      "step": 13865
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00015230386052303858,
+      "loss": 0.1578,
+      "step": 13866
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015229140722291406,
+      "loss": 0.1456,
+      "step": 13867
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015227895392278952,
+      "loss": 0.1519,
+      "step": 13868
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.000152266500622665,
+      "loss": 0.1733,
+      "step": 13869
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015225404732254044,
+      "loss": 0.5122,
+      "step": 13870
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015224159402241593,
+      "loss": 0.3487,
+      "step": 13871
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015222914072229139,
+      "loss": 0.2392,
+      "step": 13872
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015221668742216687,
+      "loss": 0.1554,
+      "step": 13873
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015220423412204233,
+      "loss": 0.2587,
+      "step": 13874
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015219178082191782,
+      "loss": 0.2516,
+      "step": 13875
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015217932752179325,
+      "loss": 0.2247,
+      "step": 13876
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015216687422166874,
+      "loss": 0.3558,
+      "step": 13877
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0001521544209215442,
+      "loss": 0.3079,
+      "step": 13878
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015214196762141965,
+      "loss": 0.267,
+      "step": 13879
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015212951432129514,
+      "loss": 0.1948,
+      "step": 13880
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015211706102117057,
+      "loss": 0.1667,
+      "step": 13881
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015210460772104606,
+      "loss": 0.1458,
+      "step": 13882
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015209215442092152,
+      "loss": 0.2641,
+      "step": 13883
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.000152079701120797,
+      "loss": 0.3372,
+      "step": 13884
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015206724782067246,
+      "loss": 0.1453,
+      "step": 13885
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015205479452054795,
+      "loss": 0.1465,
+      "step": 13886
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015204234122042338,
+      "loss": 0.1937,
+      "step": 13887
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015202988792029887,
+      "loss": 0.1758,
+      "step": 13888
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015201743462017433,
+      "loss": 0.3096,
+      "step": 13889
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015200498132004981,
+      "loss": 0.211,
+      "step": 13890
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015199252801992527,
+      "loss": 0.2121,
+      "step": 13891
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00015198007471980076,
+      "loss": 0.2866,
+      "step": 13892
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0001519676214196762,
+      "loss": 0.1208,
+      "step": 13893
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015195516811955165,
+      "loss": 0.1958,
+      "step": 13894
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015194271481942714,
+      "loss": 0.2749,
+      "step": 13895
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015194271481942714,
+      "loss": 0.5479,
+      "step": 13896
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0001519302615193026,
+      "loss": 0.2193,
+      "step": 13897
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015191780821917808,
+      "loss": 0.189,
+      "step": 13898
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015190535491905351,
+      "loss": 0.3054,
+      "step": 13899
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.000151892901618929,
+      "loss": 0.1836,
+      "step": 13900
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015188044831880446,
+      "loss": 0.1852,
+      "step": 13901
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015186799501867995,
+      "loss": 0.3443,
+      "step": 13902
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0001518555417185554,
+      "loss": 0.3338,
+      "step": 13903
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0001518430884184309,
+      "loss": 0.1745,
+      "step": 13904
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015183063511830632,
+      "loss": 0.1387,
+      "step": 13905
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0001518181818181818,
+      "loss": 0.2477,
+      "step": 13906
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015180572851805727,
+      "loss": 0.2647,
+      "step": 13907
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015179327521793276,
+      "loss": 0.3807,
+      "step": 13908
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015178082191780821,
+      "loss": 0.2053,
+      "step": 13909
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0001517683686176837,
+      "loss": 0.1992,
+      "step": 13910
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015175591531755913,
+      "loss": 0.1823,
+      "step": 13911
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0001517434620174346,
+      "loss": 0.2221,
+      "step": 13912
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015173100871731008,
+      "loss": 0.16,
+      "step": 13913
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015171855541718554,
+      "loss": 0.2213,
+      "step": 13914
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015170610211706102,
+      "loss": 0.216,
+      "step": 13915
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015169364881693646,
+      "loss": 0.1694,
+      "step": 13916
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00015168119551681194,
+      "loss": 0.1776,
+      "step": 13917
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0001516687422166874,
+      "loss": 0.2197,
+      "step": 13918
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0001516562889165629,
+      "loss": 0.1481,
+      "step": 13919
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015164383561643835,
+      "loss": 0.1609,
+      "step": 13920
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015163138231631383,
+      "loss": 0.3164,
+      "step": 13921
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015161892901618926,
+      "loss": 0.1288,
+      "step": 13922
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015160647571606475,
+      "loss": 0.1807,
+      "step": 13923
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0001515940224159402,
+      "loss": 0.1394,
+      "step": 13924
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0001515815691158157,
+      "loss": 0.1503,
+      "step": 13925
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015156911581569116,
+      "loss": 0.1949,
+      "step": 13926
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0001515566625155666,
+      "loss": 0.2133,
+      "step": 13927
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015154420921544207,
+      "loss": 0.1654,
+      "step": 13928
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015153175591531753,
+      "loss": 0.1936,
+      "step": 13929
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015151930261519302,
+      "loss": 0.1495,
+      "step": 13930
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015150684931506848,
+      "loss": 0.1355,
+      "step": 13931
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015149439601494397,
+      "loss": 0.2334,
+      "step": 13932
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0001514819427148194,
+      "loss": 0.2883,
+      "step": 13933
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015146948941469488,
+      "loss": 0.1569,
+      "step": 13934
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015145703611457034,
+      "loss": 0.1559,
+      "step": 13935
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015144458281444583,
+      "loss": 0.2952,
+      "step": 13936
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0001514321295143213,
+      "loss": 0.1886,
+      "step": 13937
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015141967621419677,
+      "loss": 0.2637,
+      "step": 13938
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0001514072229140722,
+      "loss": 0.2416,
+      "step": 13939
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0001513947696139477,
+      "loss": 0.1731,
+      "step": 13940
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015138231631382315,
+      "loss": 0.1659,
+      "step": 13941
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015136986301369864,
+      "loss": 0.1622,
+      "step": 13942
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0001513574097135741,
+      "loss": 0.1742,
+      "step": 13943
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015134495641344953,
+      "loss": 0.208,
+      "step": 13944
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.00015133250311332502,
+      "loss": 0.1961,
+      "step": 13945
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015132004981320047,
+      "loss": 0.1888,
+      "step": 13946
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015130759651307596,
+      "loss": 0.3019,
+      "step": 13947
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015129514321295142,
+      "loss": 0.1863,
+      "step": 13948
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0001512826899128269,
+      "loss": 0.1572,
+      "step": 13949
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015127023661270234,
+      "loss": 0.2208,
+      "step": 13950
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015125778331257782,
+      "loss": 0.254,
+      "step": 13951
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015124533001245328,
+      "loss": 0.2232,
+      "step": 13952
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015123287671232877,
+      "loss": 0.3522,
+      "step": 13953
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015122042341220423,
+      "loss": 0.3727,
+      "step": 13954
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0001512079701120797,
+      "loss": 0.2156,
+      "step": 13955
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015119551681195515,
+      "loss": 0.1436,
+      "step": 13956
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015118306351183063,
+      "loss": 0.1688,
+      "step": 13957
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0001511706102117061,
+      "loss": 0.3577,
+      "step": 13958
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015115815691158155,
+      "loss": 0.1941,
+      "step": 13959
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015114570361145704,
+      "loss": 0.1712,
+      "step": 13960
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015113325031133247,
+      "loss": 0.1645,
+      "step": 13961
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015112079701120796,
+      "loss": 0.1706,
+      "step": 13962
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015110834371108342,
+      "loss": 0.326,
+      "step": 13963
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0001510958904109589,
+      "loss": 0.2187,
+      "step": 13964
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015108343711083436,
+      "loss": 0.2965,
+      "step": 13965
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015107098381070982,
+      "loss": 0.2032,
+      "step": 13966
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015105853051058528,
+      "loss": 0.28,
+      "step": 13967
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015104607721046077,
+      "loss": 0.1461,
+      "step": 13968
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015103362391033623,
+      "loss": 0.1912,
+      "step": 13969
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0001510211706102117,
+      "loss": 0.2797,
+      "step": 13970
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00015100871731008717,
+      "loss": 0.1734,
+      "step": 13971
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015099626400996263,
+      "loss": 0.181,
+      "step": 13972
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0001509838107098381,
+      "loss": 0.1413,
+      "step": 13973
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015097135740971355,
+      "loss": 0.2285,
+      "step": 13974
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015095890410958903,
+      "loss": 0.1334,
+      "step": 13975
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0001509464508094645,
+      "loss": 0.2273,
+      "step": 13976
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015093399750933995,
+      "loss": 0.222,
+      "step": 13977
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0001509215442092154,
+      "loss": 0.1667,
+      "step": 13978
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0001509090909090909,
+      "loss": 0.1921,
+      "step": 13979
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015089663760896636,
+      "loss": 0.1387,
+      "step": 13980
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015088418430884184,
+      "loss": 0.1726,
+      "step": 13981
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015087173100871728,
+      "loss": 0.1345,
+      "step": 13982
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015085927770859276,
+      "loss": 0.2125,
+      "step": 13983
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015084682440846822,
+      "loss": 0.2272,
+      "step": 13984
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0001508343711083437,
+      "loss": 0.3598,
+      "step": 13985
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015082191780821917,
+      "loss": 0.285,
+      "step": 13986
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015080946450809465,
+      "loss": 0.1534,
+      "step": 13987
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015079701120797009,
+      "loss": 0.1498,
+      "step": 13988
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015078455790784557,
+      "loss": 0.2509,
+      "step": 13989
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015077210460772103,
+      "loss": 0.2822,
+      "step": 13990
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0001507596513075965,
+      "loss": 0.206,
+      "step": 13991
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015074719800747198,
+      "loss": 0.2398,
+      "step": 13992
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0001507347447073474,
+      "loss": 0.2244,
+      "step": 13993
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0001507222914072229,
+      "loss": 0.2094,
+      "step": 13994
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015070983810709835,
+      "loss": 0.2894,
+      "step": 13995
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00015069738480697384,
+      "loss": 0.2445,
+      "step": 13996
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0001506849315068493,
+      "loss": 0.1598,
+      "step": 13997
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015067247820672479,
+      "loss": 0.2786,
+      "step": 13998
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015066002490660022,
+      "loss": 0.1697,
+      "step": 13999
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0001506475716064757,
+      "loss": 0.1645,
+      "step": 14000
+    },
+    {
+      "epoch": 5.37,
+      "eval_cer": 0.10441500673078799,
+      "eval_loss": 0.35686662793159485,
+      "eval_runtime": 46.7314,
+      "eval_samples_per_second": 37.512,
+      "eval_steps_per_second": 4.708,
+      "eval_wer": 0.17129022039190842,
+      "step": 14000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015063511830635116,
+      "loss": 0.1533,
+      "step": 14001
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015062266500622665,
+      "loss": 0.2354,
+      "step": 14002
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0001506102117061021,
+      "loss": 0.1912,
+      "step": 14003
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0001505977584059776,
+      "loss": 0.1772,
+      "step": 14004
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015058530510585303,
+      "loss": 0.2835,
+      "step": 14005
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015057285180572849,
+      "loss": 0.1862,
+      "step": 14006
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015056039850560397,
+      "loss": 0.139,
+      "step": 14007
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015054794520547943,
+      "loss": 0.3105,
+      "step": 14008
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015053549190535492,
+      "loss": 0.1743,
+      "step": 14009
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015052303860523035,
+      "loss": 0.1561,
+      "step": 14010
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015051058530510584,
+      "loss": 0.1514,
+      "step": 14011
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0001504981320049813,
+      "loss": 0.1616,
+      "step": 14012
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015048567870485678,
+      "loss": 0.1204,
+      "step": 14013
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015047322540473224,
+      "loss": 0.3248,
+      "step": 14014
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015046077210460773,
+      "loss": 0.2056,
+      "step": 14015
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015044831880448316,
+      "loss": 0.2643,
+      "step": 14016
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015043586550435864,
+      "loss": 0.1123,
+      "step": 14017
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0001504234122042341,
+      "loss": 0.1637,
+      "step": 14018
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0001504109589041096,
+      "loss": 0.2068,
+      "step": 14019
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015039850560398505,
+      "loss": 0.3146,
+      "step": 14020
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015038605230386048,
+      "loss": 0.1736,
+      "step": 14021
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015037359900373597,
+      "loss": 0.1756,
+      "step": 14022
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00015036114570361143,
+      "loss": 0.1313,
+      "step": 14023
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0001503486924034869,
+      "loss": 0.2474,
+      "step": 14024
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015033623910336237,
+      "loss": 0.0866,
+      "step": 14025
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015032378580323786,
+      "loss": 0.2659,
+      "step": 14026
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0001503113325031133,
+      "loss": 0.2653,
+      "step": 14027
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015029887920298878,
+      "loss": 0.199,
+      "step": 14028
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015028642590286424,
+      "loss": 0.2688,
+      "step": 14029
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015027397260273972,
+      "loss": 0.1679,
+      "step": 14030
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015026151930261518,
+      "loss": 0.1929,
+      "step": 14031
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015024906600249067,
+      "loss": 0.1994,
+      "step": 14032
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0001502366127023661,
+      "loss": 0.3271,
+      "step": 14033
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015022415940224159,
+      "loss": 0.1933,
+      "step": 14034
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015021170610211705,
+      "loss": 0.2815,
+      "step": 14035
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015019925280199253,
+      "loss": 0.1299,
+      "step": 14036
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.000150186799501868,
+      "loss": 0.1512,
+      "step": 14037
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015017434620174342,
+      "loss": 0.2057,
+      "step": 14038
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0001501618929016189,
+      "loss": 0.2588,
+      "step": 14039
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015014943960149437,
+      "loss": 0.2056,
+      "step": 14040
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015013698630136985,
+      "loss": 0.1789,
+      "step": 14041
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015012453300124531,
+      "loss": 0.2167,
+      "step": 14042
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0001501120797011208,
+      "loss": 0.1572,
+      "step": 14043
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015009962640099623,
+      "loss": 0.2697,
+      "step": 14044
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015008717310087172,
+      "loss": 0.2582,
+      "step": 14045
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015007471980074718,
+      "loss": 0.2719,
+      "step": 14046
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015006226650062266,
+      "loss": 0.1907,
+      "step": 14047
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00015004981320049812,
+      "loss": 0.2226,
+      "step": 14048
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0001500373599003736,
+      "loss": 0.1879,
+      "step": 14049
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00015002490660024904,
+      "loss": 0.1333,
+      "step": 14050
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00015001245330012453,
+      "loss": 0.1778,
+      "step": 14051
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00015,
+      "loss": 0.2363,
+      "step": 14052
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014998754669987545,
+      "loss": 0.1833,
+      "step": 14053
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014997509339975093,
+      "loss": 0.2475,
+      "step": 14054
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0001499626400996264,
+      "loss": 0.1795,
+      "step": 14055
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014995018679950185,
+      "loss": 0.1932,
+      "step": 14056
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014993773349937734,
+      "loss": 0.1723,
+      "step": 14057
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0001499252801992528,
+      "loss": 0.2232,
+      "step": 14058
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014991282689912826,
+      "loss": 0.267,
+      "step": 14059
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014990037359900371,
+      "loss": 0.1941,
+      "step": 14060
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014988792029887917,
+      "loss": 0.1368,
+      "step": 14061
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014987546699875466,
+      "loss": 0.2218,
+      "step": 14062
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014986301369863012,
+      "loss": 0.1674,
+      "step": 14063
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014985056039850558,
+      "loss": 0.196,
+      "step": 14064
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014983810709838106,
+      "loss": 0.1928,
+      "step": 14065
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014982565379825652,
+      "loss": 0.1678,
+      "step": 14066
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014981320049813198,
+      "loss": 0.2254,
+      "step": 14067
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014980074719800747,
+      "loss": 0.192,
+      "step": 14068
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014978829389788293,
+      "loss": 0.1842,
+      "step": 14069
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0001497758405977584,
+      "loss": 0.2132,
+      "step": 14070
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014976338729763387,
+      "loss": 0.2151,
+      "step": 14071
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014975093399750933,
+      "loss": 0.1817,
+      "step": 14072
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0001497384806973848,
+      "loss": 0.1594,
+      "step": 14073
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014972602739726028,
+      "loss": 0.1656,
+      "step": 14074
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00014971357409713574,
+      "loss": 0.1862,
+      "step": 14075
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0001497011207970112,
+      "loss": 0.1895,
+      "step": 14076
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014968866749688666,
+      "loss": 0.2452,
+      "step": 14077
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014967621419676211,
+      "loss": 0.1371,
+      "step": 14078
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0001496637608966376,
+      "loss": 0.166,
+      "step": 14079
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014965130759651306,
+      "loss": 0.1615,
+      "step": 14080
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014963885429638852,
+      "loss": 0.2224,
+      "step": 14081
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.000149626400996264,
+      "loss": 0.1893,
+      "step": 14082
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014961394769613947,
+      "loss": 0.3889,
+      "step": 14083
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014960149439601492,
+      "loss": 0.1398,
+      "step": 14084
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0001495890410958904,
+      "loss": 0.1944,
+      "step": 14085
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014957658779576587,
+      "loss": 0.1672,
+      "step": 14086
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014956413449564133,
+      "loss": 0.1229,
+      "step": 14087
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014955168119551682,
+      "loss": 0.192,
+      "step": 14088
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014953922789539227,
+      "loss": 0.3485,
+      "step": 14089
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014952677459526773,
+      "loss": 0.3004,
+      "step": 14090
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014951432129514322,
+      "loss": 0.2514,
+      "step": 14091
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014950186799501865,
+      "loss": 0.1604,
+      "step": 14092
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014948941469489414,
+      "loss": 0.1378,
+      "step": 14093
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0001494769613947696,
+      "loss": 0.1472,
+      "step": 14094
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014946450809464506,
+      "loss": 0.3972,
+      "step": 14095
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014945205479452054,
+      "loss": 0.2068,
+      "step": 14096
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.000149439601494396,
+      "loss": 0.2895,
+      "step": 14097
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014942714819427146,
+      "loss": 0.1409,
+      "step": 14098
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014941469489414695,
+      "loss": 0.1591,
+      "step": 14099
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0001494022415940224,
+      "loss": 0.1527,
+      "step": 14100
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00014938978829389787,
+      "loss": 0.2354,
+      "step": 14101
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014937733499377335,
+      "loss": 0.2682,
+      "step": 14102
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0001493648816936488,
+      "loss": 0.1903,
+      "step": 14103
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014935242839352427,
+      "loss": 0.1715,
+      "step": 14104
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014933997509339976,
+      "loss": 0.1793,
+      "step": 14105
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014932752179327522,
+      "loss": 0.1555,
+      "step": 14106
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014931506849315067,
+      "loss": 0.1569,
+      "step": 14107
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014930261519302613,
+      "loss": 0.3182,
+      "step": 14108
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0001492901618929016,
+      "loss": 0.1642,
+      "step": 14109
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014927770859277708,
+      "loss": 0.3166,
+      "step": 14110
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014926525529265254,
+      "loss": 0.1183,
+      "step": 14111
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.000149252801992528,
+      "loss": 0.2528,
+      "step": 14112
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014924034869240348,
+      "loss": 0.244,
+      "step": 14113
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014922789539227894,
+      "loss": 0.2258,
+      "step": 14114
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0001492154420921544,
+      "loss": 0.1662,
+      "step": 14115
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0001492029887920299,
+      "loss": 0.3587,
+      "step": 14116
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014919053549190535,
+      "loss": 0.1788,
+      "step": 14117
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0001491780821917808,
+      "loss": 0.1802,
+      "step": 14118
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0001491656288916563,
+      "loss": 0.1288,
+      "step": 14119
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014915317559153175,
+      "loss": 0.2833,
+      "step": 14120
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0001491407222914072,
+      "loss": 0.1755,
+      "step": 14121
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0001491282689912827,
+      "loss": 0.2139,
+      "step": 14122
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014911581569115813,
+      "loss": 0.2076,
+      "step": 14123
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014910336239103362,
+      "loss": 0.1341,
+      "step": 14124
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014909090909090908,
+      "loss": 0.1956,
+      "step": 14125
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014907845579078453,
+      "loss": 0.1721,
+      "step": 14126
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00014906600249066002,
+      "loss": 0.2152,
+      "step": 14127
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014905354919053548,
+      "loss": 0.2965,
+      "step": 14128
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014904109589041094,
+      "loss": 0.1659,
+      "step": 14129
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014902864259028643,
+      "loss": 0.1434,
+      "step": 14130
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014901618929016188,
+      "loss": 0.2531,
+      "step": 14131
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014900373599003734,
+      "loss": 0.239,
+      "step": 14132
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0001489912826899128,
+      "loss": 0.1857,
+      "step": 14133
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0001489788293897883,
+      "loss": 0.1939,
+      "step": 14134
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014896637608966375,
+      "loss": 0.183,
+      "step": 14135
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0001489539227895392,
+      "loss": 0.1662,
+      "step": 14136
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0001489414694894147,
+      "loss": 0.1866,
+      "step": 14137
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014892901618929015,
+      "loss": 0.2477,
+      "step": 14138
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0001489165628891656,
+      "loss": 0.3358,
+      "step": 14139
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014890410958904107,
+      "loss": 0.2208,
+      "step": 14140
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014889165628891656,
+      "loss": 0.1586,
+      "step": 14141
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014887920298879202,
+      "loss": 0.1817,
+      "step": 14142
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014886674968866748,
+      "loss": 0.1435,
+      "step": 14143
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014885429638854294,
+      "loss": 0.1883,
+      "step": 14144
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014884184308841842,
+      "loss": 0.301,
+      "step": 14145
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014882938978829388,
+      "loss": 0.2124,
+      "step": 14146
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014881693648816934,
+      "loss": 0.3153,
+      "step": 14147
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014880448318804483,
+      "loss": 0.1837,
+      "step": 14148
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014879202988792029,
+      "loss": 0.2337,
+      "step": 14149
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014877957658779574,
+      "loss": 0.2388,
+      "step": 14150
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014876712328767123,
+      "loss": 0.2639,
+      "step": 14151
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0001487546699875467,
+      "loss": 0.2302,
+      "step": 14152
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00014874221668742215,
+      "loss": 0.3779,
+      "step": 14153
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014872976338729764,
+      "loss": 0.2188,
+      "step": 14154
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014871731008717307,
+      "loss": 0.2386,
+      "step": 14155
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014870485678704855,
+      "loss": 0.1478,
+      "step": 14156
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.000148692403486924,
+      "loss": 0.1507,
+      "step": 14157
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014867995018679947,
+      "loss": 0.2438,
+      "step": 14158
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014866749688667496,
+      "loss": 0.189,
+      "step": 14159
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014865504358655042,
+      "loss": 0.1402,
+      "step": 14160
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014864259028642588,
+      "loss": 0.1447,
+      "step": 14161
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014863013698630136,
+      "loss": 0.1597,
+      "step": 14162
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014861768368617682,
+      "loss": 0.1695,
+      "step": 14163
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014860523038605228,
+      "loss": 0.2389,
+      "step": 14164
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014859277708592777,
+      "loss": 0.2053,
+      "step": 14165
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014858032378580323,
+      "loss": 0.2534,
+      "step": 14166
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014856787048567869,
+      "loss": 0.1654,
+      "step": 14167
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014855541718555417,
+      "loss": 0.166,
+      "step": 14168
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014854296388542963,
+      "loss": 0.1542,
+      "step": 14169
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0001485305105853051,
+      "loss": 0.2073,
+      "step": 14170
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014851805728518055,
+      "loss": 0.1625,
+      "step": 14171
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.000148505603985056,
+      "loss": 0.2492,
+      "step": 14172
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0001484931506849315,
+      "loss": 0.1953,
+      "step": 14173
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014848069738480695,
+      "loss": 0.1367,
+      "step": 14174
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0001484682440846824,
+      "loss": 0.197,
+      "step": 14175
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0001484557907845579,
+      "loss": 0.2119,
+      "step": 14176
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014844333748443336,
+      "loss": 0.2734,
+      "step": 14177
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00014843088418430882,
+      "loss": 0.2141,
+      "step": 14178
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0001484184308841843,
+      "loss": 0.2225,
+      "step": 14179
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014840597758405976,
+      "loss": 0.2152,
+      "step": 14180
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014839352428393522,
+      "loss": 0.1652,
+      "step": 14181
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0001483810709838107,
+      "loss": 0.1485,
+      "step": 14182
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014836861768368617,
+      "loss": 0.2884,
+      "step": 14183
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014835616438356163,
+      "loss": 0.2369,
+      "step": 14184
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0001483437110834371,
+      "loss": 0.1746,
+      "step": 14185
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014833125778331255,
+      "loss": 0.2272,
+      "step": 14186
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014831880448318803,
+      "loss": 0.2025,
+      "step": 14187
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0001483063511830635,
+      "loss": 0.1686,
+      "step": 14188
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014829389788293895,
+      "loss": 0.2773,
+      "step": 14189
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014828144458281444,
+      "loss": 0.1917,
+      "step": 14190
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0001482689912826899,
+      "loss": 0.1914,
+      "step": 14191
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014825653798256535,
+      "loss": 0.2265,
+      "step": 14192
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014824408468244084,
+      "loss": 0.1313,
+      "step": 14193
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0001482316313823163,
+      "loss": 0.1897,
+      "step": 14194
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014821917808219176,
+      "loss": 0.4032,
+      "step": 14195
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014820672478206725,
+      "loss": 0.1991,
+      "step": 14196
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0001481942714819427,
+      "loss": 0.1882,
+      "step": 14197
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014818181818181816,
+      "loss": 0.2297,
+      "step": 14198
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014816936488169365,
+      "loss": 0.1822,
+      "step": 14199
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0001481569115815691,
+      "loss": 0.2011,
+      "step": 14200
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014814445828144457,
+      "loss": 0.175,
+      "step": 14201
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014813200498132003,
+      "loss": 0.3439,
+      "step": 14202
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0001481195516811955,
+      "loss": 0.1593,
+      "step": 14203
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014810709838107097,
+      "loss": 0.1947,
+      "step": 14204
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00014809464508094643,
+      "loss": 0.2904,
+      "step": 14205
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0001480821917808219,
+      "loss": 0.2653,
+      "step": 14206
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014806973848069738,
+      "loss": 0.1875,
+      "step": 14207
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014805728518057284,
+      "loss": 0.3068,
+      "step": 14208
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0001480448318804483,
+      "loss": 0.2154,
+      "step": 14209
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014803237858032378,
+      "loss": 0.2294,
+      "step": 14210
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014801992528019924,
+      "loss": 0.1545,
+      "step": 14211
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0001480074719800747,
+      "loss": 0.1387,
+      "step": 14212
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0001479950186799502,
+      "loss": 0.1342,
+      "step": 14213
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014798256537982565,
+      "loss": 0.2401,
+      "step": 14214
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0001479701120797011,
+      "loss": 0.3337,
+      "step": 14215
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0001479576587795766,
+      "loss": 0.2025,
+      "step": 14216
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014794520547945202,
+      "loss": 0.2135,
+      "step": 14217
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0001479327521793275,
+      "loss": 0.1669,
+      "step": 14218
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014792029887920297,
+      "loss": 0.1737,
+      "step": 14219
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014790784557907843,
+      "loss": 0.2228,
+      "step": 14220
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014789539227895391,
+      "loss": 0.2642,
+      "step": 14221
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014788293897882937,
+      "loss": 0.1793,
+      "step": 14222
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014787048567870483,
+      "loss": 0.1627,
+      "step": 14223
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014785803237858032,
+      "loss": 0.1204,
+      "step": 14224
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014784557907845578,
+      "loss": 0.221,
+      "step": 14225
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014783312577833124,
+      "loss": 0.2387,
+      "step": 14226
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014782067247820672,
+      "loss": 0.2648,
+      "step": 14227
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014780821917808218,
+      "loss": 0.2602,
+      "step": 14228
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014779576587795764,
+      "loss": 0.2829,
+      "step": 14229
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014778331257783313,
+      "loss": 0.1407,
+      "step": 14230
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0001477708592777086,
+      "loss": 0.135,
+      "step": 14231
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00014775840597758405,
+      "loss": 0.2055,
+      "step": 14232
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0001477459526774595,
+      "loss": 0.1947,
+      "step": 14233
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014773349937733497,
+      "loss": 0.2362,
+      "step": 14234
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014772104607721045,
+      "loss": 0.2856,
+      "step": 14235
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0001477085927770859,
+      "loss": 0.2274,
+      "step": 14236
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014769613947696137,
+      "loss": 0.1563,
+      "step": 14237
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014768368617683686,
+      "loss": 0.1854,
+      "step": 14238
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014767123287671232,
+      "loss": 0.2696,
+      "step": 14239
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014765877957658777,
+      "loss": 0.3174,
+      "step": 14240
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014764632627646326,
+      "loss": 0.1693,
+      "step": 14241
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014763387297633872,
+      "loss": 0.1994,
+      "step": 14242
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014762141967621418,
+      "loss": 0.2263,
+      "step": 14243
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014760896637608967,
+      "loss": 0.1613,
+      "step": 14244
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014759651307596512,
+      "loss": 0.2982,
+      "step": 14245
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014758405977584058,
+      "loss": 0.1855,
+      "step": 14246
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014757160647571607,
+      "loss": 0.1944,
+      "step": 14247
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014755915317559153,
+      "loss": 0.1631,
+      "step": 14248
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.000147546699875467,
+      "loss": 0.2054,
+      "step": 14249
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014753424657534245,
+      "loss": 0.1703,
+      "step": 14250
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0001475217932752179,
+      "loss": 0.2513,
+      "step": 14251
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0001475093399750934,
+      "loss": 0.3888,
+      "step": 14252
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014749688667496885,
+      "loss": 0.2429,
+      "step": 14253
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0001474844333748443,
+      "loss": 0.234,
+      "step": 14254
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0001474719800747198,
+      "loss": 0.1704,
+      "step": 14255
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014745952677459526,
+      "loss": 0.263,
+      "step": 14256
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00014744707347447072,
+      "loss": 0.1648,
+      "step": 14257
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0001474346201743462,
+      "loss": 0.2221,
+      "step": 14258
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014742216687422166,
+      "loss": 0.1931,
+      "step": 14259
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014740971357409712,
+      "loss": 0.1746,
+      "step": 14260
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0001473972602739726,
+      "loss": 0.158,
+      "step": 14261
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014738480697384807,
+      "loss": 0.1163,
+      "step": 14262
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014737235367372352,
+      "loss": 0.2358,
+      "step": 14263
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.000147359900373599,
+      "loss": 0.2157,
+      "step": 14264
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014734744707347444,
+      "loss": 0.2252,
+      "step": 14265
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014733499377334993,
+      "loss": 0.1423,
+      "step": 14266
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0001473225404732254,
+      "loss": 0.1105,
+      "step": 14267
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014731008717310085,
+      "loss": 0.1245,
+      "step": 14268
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014729763387297633,
+      "loss": 0.1402,
+      "step": 14269
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0001472851805728518,
+      "loss": 0.4054,
+      "step": 14270
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014727272727272725,
+      "loss": 0.2359,
+      "step": 14271
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014726027397260274,
+      "loss": 0.2091,
+      "step": 14272
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0001472478206724782,
+      "loss": 0.217,
+      "step": 14273
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014723536737235366,
+      "loss": 0.212,
+      "step": 14274
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014722291407222914,
+      "loss": 0.2136,
+      "step": 14275
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0001472104607721046,
+      "loss": 0.1948,
+      "step": 14276
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014719800747198006,
+      "loss": 0.3252,
+      "step": 14277
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014718555417185555,
+      "loss": 0.1766,
+      "step": 14278
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.000147173100871731,
+      "loss": 0.1574,
+      "step": 14279
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014716064757160647,
+      "loss": 0.1449,
+      "step": 14280
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014714819427148193,
+      "loss": 0.1735,
+      "step": 14281
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014713574097135738,
+      "loss": 0.2224,
+      "step": 14282
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014712328767123287,
+      "loss": 0.3114,
+      "step": 14283
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.00014711083437110833,
+      "loss": 0.2404,
+      "step": 14284
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0001470983810709838,
+      "loss": 0.1918,
+      "step": 14285
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014708592777085928,
+      "loss": 0.1298,
+      "step": 14286
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014707347447073473,
+      "loss": 0.1884,
+      "step": 14287
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0001470610211706102,
+      "loss": 0.1618,
+      "step": 14288
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014704856787048568,
+      "loss": 0.2052,
+      "step": 14289
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014703611457036114,
+      "loss": 0.2091,
+      "step": 14290
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0001470236612702366,
+      "loss": 0.1896,
+      "step": 14291
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014701120797011208,
+      "loss": 0.1601,
+      "step": 14292
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014699875466998754,
+      "loss": 0.2334,
+      "step": 14293
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.000146986301369863,
+      "loss": 0.1503,
+      "step": 14294
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014697384806973846,
+      "loss": 0.2551,
+      "step": 14295
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014696139476961392,
+      "loss": 0.2751,
+      "step": 14296
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0001469489414694894,
+      "loss": 0.1786,
+      "step": 14297
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014693648816936487,
+      "loss": 0.2246,
+      "step": 14298
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014692403486924033,
+      "loss": 0.1764,
+      "step": 14299
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0001469115815691158,
+      "loss": 0.2099,
+      "step": 14300
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014689912826899127,
+      "loss": 0.2089,
+      "step": 14301
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014688667496886673,
+      "loss": 0.2859,
+      "step": 14302
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014687422166874222,
+      "loss": 0.3709,
+      "step": 14303
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014686176836861768,
+      "loss": 0.1463,
+      "step": 14304
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014684931506849314,
+      "loss": 0.1617,
+      "step": 14305
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0001468368617683686,
+      "loss": 0.2319,
+      "step": 14306
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014682440846824408,
+      "loss": 0.2354,
+      "step": 14307
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014681195516811954,
+      "loss": 0.156,
+      "step": 14308
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.000146799501867995,
+      "loss": 0.1822,
+      "step": 14309
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.00014678704856787049,
+      "loss": 0.3356,
+      "step": 14310
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014677459526774594,
+      "loss": 0.0953,
+      "step": 14311
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0001467621419676214,
+      "loss": 0.2388,
+      "step": 14312
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014674968866749686,
+      "loss": 0.1414,
+      "step": 14313
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014673723536737235,
+      "loss": 0.2726,
+      "step": 14314
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0001467247820672478,
+      "loss": 0.2029,
+      "step": 14315
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014671232876712327,
+      "loss": 0.2681,
+      "step": 14316
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014669987546699873,
+      "loss": 0.2229,
+      "step": 14317
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0001466874221668742,
+      "loss": 0.1181,
+      "step": 14318
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014667496886674967,
+      "loss": 0.1404,
+      "step": 14319
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014666251556662513,
+      "loss": 0.2944,
+      "step": 14320
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014665006226650062,
+      "loss": 0.238,
+      "step": 14321
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014663760896637608,
+      "loss": 0.1737,
+      "step": 14322
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014662515566625154,
+      "loss": 0.1273,
+      "step": 14323
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014661270236612702,
+      "loss": 0.207,
+      "step": 14324
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014660024906600248,
+      "loss": 0.1524,
+      "step": 14325
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014658779576587794,
+      "loss": 0.1876,
+      "step": 14326
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0001465753424657534,
+      "loss": 0.3602,
+      "step": 14327
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014656288916562886,
+      "loss": 0.2225,
+      "step": 14328
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014655043586550435,
+      "loss": 0.2635,
+      "step": 14329
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0001465379825653798,
+      "loss": 0.187,
+      "step": 14330
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014652552926525526,
+      "loss": 0.1222,
+      "step": 14331
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014651307596513075,
+      "loss": 0.1961,
+      "step": 14332
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0001465006226650062,
+      "loss": 0.3672,
+      "step": 14333
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014648816936488167,
+      "loss": 0.1706,
+      "step": 14334
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00014647571606475715,
+      "loss": 0.1924,
+      "step": 14335
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0001464632627646326,
+      "loss": 0.1689,
+      "step": 14336
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014645080946450807,
+      "loss": 0.2124,
+      "step": 14337
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014643835616438356,
+      "loss": 0.2818,
+      "step": 14338
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014642590286425902,
+      "loss": 0.33,
+      "step": 14339
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014641344956413448,
+      "loss": 0.1777,
+      "step": 14340
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014640099626400996,
+      "loss": 0.2304,
+      "step": 14341
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014638854296388542,
+      "loss": 0.2217,
+      "step": 14342
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014637608966376088,
+      "loss": 0.162,
+      "step": 14343
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014636363636363634,
+      "loss": 0.2851,
+      "step": 14344
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0001463511830635118,
+      "loss": 0.2285,
+      "step": 14345
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014633872976338729,
+      "loss": 0.1999,
+      "step": 14346
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014632627646326275,
+      "loss": 0.2302,
+      "step": 14347
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0001463138231631382,
+      "loss": 0.1333,
+      "step": 14348
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0001463013698630137,
+      "loss": 0.2855,
+      "step": 14349
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014628891656288915,
+      "loss": 0.1078,
+      "step": 14350
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0001462764632627646,
+      "loss": 0.1511,
+      "step": 14351
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0001462640099626401,
+      "loss": 0.3251,
+      "step": 14352
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014625155666251555,
+      "loss": 0.1947,
+      "step": 14353
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014623910336239101,
+      "loss": 0.152,
+      "step": 14354
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0001462266500622665,
+      "loss": 0.1211,
+      "step": 14355
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014621419676214196,
+      "loss": 0.1901,
+      "step": 14356
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014620174346201742,
+      "loss": 0.1262,
+      "step": 14357
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0001461892901618929,
+      "loss": 0.2219,
+      "step": 14358
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014617683686176834,
+      "loss": 0.1653,
+      "step": 14359
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014616438356164382,
+      "loss": 0.1398,
+      "step": 14360
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014615193026151928,
+      "loss": 0.1583,
+      "step": 14361
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00014613947696139474,
+      "loss": 0.2417,
+      "step": 14362
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014612702366127023,
+      "loss": 0.1098,
+      "step": 14363
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0001461145703611457,
+      "loss": 0.4127,
+      "step": 14364
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014610211706102115,
+      "loss": 0.1625,
+      "step": 14365
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014608966376089663,
+      "loss": 0.1942,
+      "step": 14366
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0001460772104607721,
+      "loss": 0.2442,
+      "step": 14367
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014606475716064755,
+      "loss": 0.245,
+      "step": 14368
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014605230386052304,
+      "loss": 0.2287,
+      "step": 14369
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0001460398505603985,
+      "loss": 0.2286,
+      "step": 14370
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014602739726027396,
+      "loss": 0.2691,
+      "step": 14371
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014601494396014944,
+      "loss": 0.1559,
+      "step": 14372
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0001460024906600249,
+      "loss": 0.2441,
+      "step": 14373
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014599003735990036,
+      "loss": 0.1581,
+      "step": 14374
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014597758405977582,
+      "loss": 0.1554,
+      "step": 14375
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014596513075965128,
+      "loss": 0.1963,
+      "step": 14376
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014595267745952676,
+      "loss": 0.3042,
+      "step": 14377
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014594022415940222,
+      "loss": 0.1617,
+      "step": 14378
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014592777085927768,
+      "loss": 0.1861,
+      "step": 14379
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014591531755915317,
+      "loss": 0.1341,
+      "step": 14380
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014590286425902863,
+      "loss": 0.1902,
+      "step": 14381
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0001458904109589041,
+      "loss": 0.2126,
+      "step": 14382
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014587795765877957,
+      "loss": 0.2389,
+      "step": 14383
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014586550435865503,
+      "loss": 0.1861,
+      "step": 14384
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0001458530510585305,
+      "loss": 0.2166,
+      "step": 14385
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014584059775840598,
+      "loss": 0.1879,
+      "step": 14386
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.00014582814445828144,
+      "loss": 0.2317,
+      "step": 14387
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0001458156911581569,
+      "loss": 0.2101,
+      "step": 14388
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014580323785803238,
+      "loss": 0.3426,
+      "step": 14389
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014579078455790782,
+      "loss": 0.2678,
+      "step": 14390
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0001457783312577833,
+      "loss": 0.2267,
+      "step": 14391
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014576587795765876,
+      "loss": 0.17,
+      "step": 14392
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014575342465753422,
+      "loss": 0.3564,
+      "step": 14393
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0001457409713574097,
+      "loss": 0.1948,
+      "step": 14394
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014572851805728517,
+      "loss": 0.1907,
+      "step": 14395
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014571606475716062,
+      "loss": 0.1838,
+      "step": 14396
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0001457036114570361,
+      "loss": 0.1943,
+      "step": 14397
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014569115815691157,
+      "loss": 0.1435,
+      "step": 14398
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014567870485678703,
+      "loss": 0.1259,
+      "step": 14399
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014566625155666252,
+      "loss": 0.1464,
+      "step": 14400
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014565379825653797,
+      "loss": 0.2802,
+      "step": 14401
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014564134495641343,
+      "loss": 0.2917,
+      "step": 14402
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014562889165628892,
+      "loss": 0.252,
+      "step": 14403
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014561643835616438,
+      "loss": 0.1766,
+      "step": 14404
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014560398505603984,
+      "loss": 0.1705,
+      "step": 14405
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0001455915317559153,
+      "loss": 0.1447,
+      "step": 14406
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014557907845579076,
+      "loss": 0.1198,
+      "step": 14407
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014556662515566624,
+      "loss": 0.2796,
+      "step": 14408
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0001455541718555417,
+      "loss": 0.196,
+      "step": 14409
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014554171855541716,
+      "loss": 0.3009,
+      "step": 14410
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014552926525529265,
+      "loss": 0.1489,
+      "step": 14411
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.0001455168119551681,
+      "loss": 0.1188,
+      "step": 14412
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014550435865504357,
+      "loss": 0.1696,
+      "step": 14413
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.00014549190535491905,
+      "loss": 0.3354,
+      "step": 14414
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0001454794520547945,
+      "loss": 0.2421,
+      "step": 14415
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014546699875466997,
+      "loss": 0.162,
+      "step": 14416
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014545454545454546,
+      "loss": 0.1787,
+      "step": 14417
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014544209215442092,
+      "loss": 0.1257,
+      "step": 14418
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014542963885429637,
+      "loss": 0.1372,
+      "step": 14419
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014541718555417186,
+      "loss": 0.2049,
+      "step": 14420
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014540473225404732,
+      "loss": 0.1883,
+      "step": 14421
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014539227895392278,
+      "loss": 0.203,
+      "step": 14422
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014537982565379824,
+      "loss": 0.1859,
+      "step": 14423
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0001453673723536737,
+      "loss": 0.1494,
+      "step": 14424
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014535491905354918,
+      "loss": 0.2098,
+      "step": 14425
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014534246575342464,
+      "loss": 0.3181,
+      "step": 14426
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0001453300124533001,
+      "loss": 0.248,
+      "step": 14427
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0001453175591531756,
+      "loss": 0.1419,
+      "step": 14428
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014530510585305105,
+      "loss": 0.19,
+      "step": 14429
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0001452926525529265,
+      "loss": 0.1517,
+      "step": 14430
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.000145280199252802,
+      "loss": 0.1879,
+      "step": 14431
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014526774595267745,
+      "loss": 0.1665,
+      "step": 14432
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0001452552926525529,
+      "loss": 0.3857,
+      "step": 14433
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0001452428393524284,
+      "loss": 0.2084,
+      "step": 14434
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014523038605230386,
+      "loss": 0.236,
+      "step": 14435
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014521793275217932,
+      "loss": 0.2236,
+      "step": 14436
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014520547945205478,
+      "loss": 0.1576,
+      "step": 14437
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014519302615193023,
+      "loss": 0.3233,
+      "step": 14438
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014518057285180572,
+      "loss": 0.3073,
+      "step": 14439
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00014516811955168118,
+      "loss": 0.2648,
+      "step": 14440
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014515566625155664,
+      "loss": 0.2177,
+      "step": 14441
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014514321295143213,
+      "loss": 0.2186,
+      "step": 14442
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014513075965130758,
+      "loss": 0.2447,
+      "step": 14443
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014511830635118304,
+      "loss": 0.15,
+      "step": 14444
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014510585305105853,
+      "loss": 0.2611,
+      "step": 14445
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.000145093399750934,
+      "loss": 0.2125,
+      "step": 14446
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014508094645080945,
+      "loss": 0.2224,
+      "step": 14447
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014506849315068493,
+      "loss": 0.1534,
+      "step": 14448
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0001450560398505604,
+      "loss": 0.147,
+      "step": 14449
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014504358655043585,
+      "loss": 0.1752,
+      "step": 14450
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014503113325031134,
+      "loss": 0.2028,
+      "step": 14451
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0001450186799501868,
+      "loss": 0.4141,
+      "step": 14452
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014500622665006226,
+      "loss": 0.3113,
+      "step": 14453
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014499377334993772,
+      "loss": 0.1609,
+      "step": 14454
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014498132004981318,
+      "loss": 0.1543,
+      "step": 14455
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014496886674968866,
+      "loss": 0.2783,
+      "step": 14456
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014495641344956412,
+      "loss": 0.2763,
+      "step": 14457
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014494396014943958,
+      "loss": 0.3142,
+      "step": 14458
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014493150684931507,
+      "loss": 0.1859,
+      "step": 14459
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014491905354919053,
+      "loss": 0.1867,
+      "step": 14460
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014490660024906599,
+      "loss": 0.3042,
+      "step": 14461
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014489414694894147,
+      "loss": 0.1449,
+      "step": 14462
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014488169364881693,
+      "loss": 0.1652,
+      "step": 14463
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0001448692403486924,
+      "loss": 0.4143,
+      "step": 14464
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014485678704856788,
+      "loss": 0.1669,
+      "step": 14465
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00014484433374844334,
+      "loss": 0.3536,
+      "step": 14466
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0001448318804483188,
+      "loss": 0.1564,
+      "step": 14467
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014481942714819425,
+      "loss": 0.082,
+      "step": 14468
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0001448069738480697,
+      "loss": 0.1748,
+      "step": 14469
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0001447945205479452,
+      "loss": 0.274,
+      "step": 14470
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014478206724782066,
+      "loss": 0.2773,
+      "step": 14471
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014476961394769612,
+      "loss": 0.2537,
+      "step": 14472
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0001447571606475716,
+      "loss": 0.1406,
+      "step": 14473
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014474470734744706,
+      "loss": 0.1567,
+      "step": 14474
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014473225404732252,
+      "loss": 0.2642,
+      "step": 14475
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.000144719800747198,
+      "loss": 0.1939,
+      "step": 14476
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014470734744707347,
+      "loss": 0.3584,
+      "step": 14477
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014469489414694893,
+      "loss": 0.2058,
+      "step": 14478
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014468244084682439,
+      "loss": 0.1625,
+      "step": 14479
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014466998754669987,
+      "loss": 0.2162,
+      "step": 14480
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014465753424657533,
+      "loss": 0.2167,
+      "step": 14481
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0001446450809464508,
+      "loss": 0.1571,
+      "step": 14482
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014463262764632628,
+      "loss": 0.2333,
+      "step": 14483
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014462017434620174,
+      "loss": 0.2297,
+      "step": 14484
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0001446077210460772,
+      "loss": 0.1414,
+      "step": 14485
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014459526774595265,
+      "loss": 0.1989,
+      "step": 14486
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0001445828144458281,
+      "loss": 0.1364,
+      "step": 14487
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0001445703611457036,
+      "loss": 0.1695,
+      "step": 14488
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014455790784557906,
+      "loss": 0.3282,
+      "step": 14489
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014454545454545452,
+      "loss": 0.1933,
+      "step": 14490
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014453300124533,
+      "loss": 0.2174,
+      "step": 14491
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00014452054794520546,
+      "loss": 0.1154,
+      "step": 14492
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014450809464508092,
+      "loss": 0.1706,
+      "step": 14493
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0001444956413449564,
+      "loss": 0.1829,
+      "step": 14494
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014448318804483187,
+      "loss": 0.1927,
+      "step": 14495
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014447073474470733,
+      "loss": 0.1889,
+      "step": 14496
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0001444582814445828,
+      "loss": 0.1696,
+      "step": 14497
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014444582814445827,
+      "loss": 0.1687,
+      "step": 14498
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014443337484433373,
+      "loss": 0.13,
+      "step": 14499
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0001444209215442092,
+      "loss": 0.2065,
+      "step": 14500
+    },
+    {
+      "epoch": 5.56,
+      "eval_cer": 0.10477788399633221,
+      "eval_loss": 0.36393919587135315,
+      "eval_runtime": 47.2045,
+      "eval_samples_per_second": 37.136,
+      "eval_steps_per_second": 4.661,
+      "eval_wer": 0.17196904557179707,
+      "step": 14500
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014440846824408465,
+      "loss": 0.2376,
+      "step": 14501
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014439601494396014,
+      "loss": 0.3162,
+      "step": 14502
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0001443835616438356,
+      "loss": 0.2169,
+      "step": 14503
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014437110834371105,
+      "loss": 0.1764,
+      "step": 14504
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014435865504358654,
+      "loss": 0.1342,
+      "step": 14505
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.000144346201743462,
+      "loss": 0.1492,
+      "step": 14506
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014433374844333746,
+      "loss": 0.2214,
+      "step": 14507
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014432129514321295,
+      "loss": 0.2869,
+      "step": 14508
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0001443088418430884,
+      "loss": 0.1977,
+      "step": 14509
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014429638854296386,
+      "loss": 0.2087,
+      "step": 14510
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014428393524283935,
+      "loss": 0.2133,
+      "step": 14511
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0001442714819427148,
+      "loss": 0.2725,
+      "step": 14512
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014425902864259027,
+      "loss": 0.2176,
+      "step": 14513
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014424657534246575,
+      "loss": 0.2757,
+      "step": 14514
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014423412204234121,
+      "loss": 0.1515,
+      "step": 14515
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014422166874221667,
+      "loss": 0.1589,
+      "step": 14516
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014420921544209213,
+      "loss": 0.1611,
+      "step": 14517
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0001441967621419676,
+      "loss": 0.1924,
+      "step": 14518
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00014418430884184308,
+      "loss": 0.1158,
+      "step": 14519
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014417185554171854,
+      "loss": 0.1694,
+      "step": 14520
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.000144159402241594,
+      "loss": 0.1819,
+      "step": 14521
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014414694894146948,
+      "loss": 0.18,
+      "step": 14522
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014413449564134494,
+      "loss": 0.3466,
+      "step": 14523
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0001441220423412204,
+      "loss": 0.2047,
+      "step": 14524
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0001441095890410959,
+      "loss": 0.1775,
+      "step": 14525
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014409713574097135,
+      "loss": 0.2039,
+      "step": 14526
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0001440846824408468,
+      "loss": 0.3139,
+      "step": 14527
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0001440722291407223,
+      "loss": 0.1822,
+      "step": 14528
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014405977584059775,
+      "loss": 0.2543,
+      "step": 14529
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0001440473225404732,
+      "loss": 0.1991,
+      "step": 14530
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014403486924034867,
+      "loss": 0.1679,
+      "step": 14531
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014402241594022413,
+      "loss": 0.1126,
+      "step": 14532
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014400996264009961,
+      "loss": 0.3217,
+      "step": 14533
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014399750933997507,
+      "loss": 0.2271,
+      "step": 14534
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014398505603985053,
+      "loss": 0.1357,
+      "step": 14535
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014397260273972602,
+      "loss": 0.12,
+      "step": 14536
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014396014943960148,
+      "loss": 0.2137,
+      "step": 14537
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014394769613947694,
+      "loss": 0.2098,
+      "step": 14538
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014393524283935242,
+      "loss": 0.2568,
+      "step": 14539
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014392278953922788,
+      "loss": 0.1616,
+      "step": 14540
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014391033623910334,
+      "loss": 0.1835,
+      "step": 14541
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014389788293897883,
+      "loss": 0.3323,
+      "step": 14542
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0001438854296388543,
+      "loss": 0.133,
+      "step": 14543
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014387297633872975,
+      "loss": 0.1356,
+      "step": 14544
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00014386052303860523,
+      "loss": 0.2393,
+      "step": 14545
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0001438480697384807,
+      "loss": 0.2147,
+      "step": 14546
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014383561643835615,
+      "loss": 0.2551,
+      "step": 14547
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0001438231631382316,
+      "loss": 0.3404,
+      "step": 14548
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014381070983810707,
+      "loss": 0.2491,
+      "step": 14549
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014379825653798256,
+      "loss": 0.186,
+      "step": 14550
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014378580323785802,
+      "loss": 0.2152,
+      "step": 14551
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014377334993773347,
+      "loss": 0.3959,
+      "step": 14552
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014376089663760896,
+      "loss": 0.1637,
+      "step": 14553
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014374844333748442,
+      "loss": 0.1604,
+      "step": 14554
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014373599003735988,
+      "loss": 0.2529,
+      "step": 14555
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014372353673723537,
+      "loss": 0.1198,
+      "step": 14556
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014371108343711082,
+      "loss": 0.2995,
+      "step": 14557
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014369863013698628,
+      "loss": 0.326,
+      "step": 14558
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014368617683686177,
+      "loss": 0.2818,
+      "step": 14559
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014367372353673723,
+      "loss": 0.1661,
+      "step": 14560
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0001436612702366127,
+      "loss": 0.1922,
+      "step": 14561
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014364881693648817,
+      "loss": 0.2107,
+      "step": 14562
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0001436363636363636,
+      "loss": 0.1552,
+      "step": 14563
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0001436239103362391,
+      "loss": 0.2351,
+      "step": 14564
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014361145703611455,
+      "loss": 0.1922,
+      "step": 14565
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014359900373599,
+      "loss": 0.2457,
+      "step": 14566
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0001435865504358655,
+      "loss": 0.1768,
+      "step": 14567
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014357409713574096,
+      "loss": 0.1947,
+      "step": 14568
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014356164383561642,
+      "loss": 0.2004,
+      "step": 14569
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0001435491905354919,
+      "loss": 0.2579,
+      "step": 14570
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00014353673723536736,
+      "loss": 0.1943,
+      "step": 14571
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014352428393524282,
+      "loss": 0.2884,
+      "step": 14572
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0001435118306351183,
+      "loss": 0.1445,
+      "step": 14573
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014349937733499377,
+      "loss": 0.2365,
+      "step": 14574
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014348692403486922,
+      "loss": 0.2482,
+      "step": 14575
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0001434744707347447,
+      "loss": 0.1593,
+      "step": 14576
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014346201743462017,
+      "loss": 0.2416,
+      "step": 14577
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014344956413449563,
+      "loss": 0.2899,
+      "step": 14578
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0001434371108343711,
+      "loss": 0.2199,
+      "step": 14579
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014342465753424655,
+      "loss": 0.1346,
+      "step": 14580
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014341220423412203,
+      "loss": 0.1434,
+      "step": 14581
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0001433997509339975,
+      "loss": 0.1954,
+      "step": 14582
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014338729763387295,
+      "loss": 0.2822,
+      "step": 14583
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014337484433374844,
+      "loss": 0.204,
+      "step": 14584
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0001433623910336239,
+      "loss": 0.1842,
+      "step": 14585
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014334993773349936,
+      "loss": 0.1246,
+      "step": 14586
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014333748443337484,
+      "loss": 0.1379,
+      "step": 14587
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0001433250311332503,
+      "loss": 0.1877,
+      "step": 14588
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014331257783312576,
+      "loss": 0.3129,
+      "step": 14589
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014330012453300125,
+      "loss": 0.1774,
+      "step": 14590
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0001432876712328767,
+      "loss": 0.1987,
+      "step": 14591
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014327521793275217,
+      "loss": 0.1848,
+      "step": 14592
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014326276463262765,
+      "loss": 0.1668,
+      "step": 14593
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014325031133250308,
+      "loss": 0.2274,
+      "step": 14594
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014323785803237857,
+      "loss": 0.2154,
+      "step": 14595
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00014322540473225403,
+      "loss": 0.2464,
+      "step": 14596
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0001432129514321295,
+      "loss": 0.1718,
+      "step": 14597
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014320049813200498,
+      "loss": 0.2261,
+      "step": 14598
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014318804483188043,
+      "loss": 0.2782,
+      "step": 14599
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0001431755915317559,
+      "loss": 0.1063,
+      "step": 14600
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014316313823163138,
+      "loss": 0.1466,
+      "step": 14601
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014315068493150684,
+      "loss": 0.3073,
+      "step": 14602
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0001431382316313823,
+      "loss": 0.1717,
+      "step": 14603
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014312577833125778,
+      "loss": 0.1509,
+      "step": 14604
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014311332503113324,
+      "loss": 0.186,
+      "step": 14605
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0001431008717310087,
+      "loss": 0.2361,
+      "step": 14606
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0001430884184308842,
+      "loss": 0.3452,
+      "step": 14607
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014307596513075965,
+      "loss": 0.3185,
+      "step": 14608
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0001430635118306351,
+      "loss": 0.2053,
+      "step": 14609
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014305105853051057,
+      "loss": 0.1995,
+      "step": 14610
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014303860523038603,
+      "loss": 0.169,
+      "step": 14611
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0001430261519302615,
+      "loss": 0.1881,
+      "step": 14612
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014301369863013697,
+      "loss": 0.1719,
+      "step": 14613
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014300124533001243,
+      "loss": 0.3803,
+      "step": 14614
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014298879202988792,
+      "loss": 0.2063,
+      "step": 14615
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014297633872976338,
+      "loss": 0.1656,
+      "step": 14616
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014296388542963884,
+      "loss": 0.1694,
+      "step": 14617
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014295143212951432,
+      "loss": 0.2201,
+      "step": 14618
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014293897882938978,
+      "loss": 0.1122,
+      "step": 14619
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014292652552926524,
+      "loss": 0.2946,
+      "step": 14620
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014291407222914073,
+      "loss": 0.1807,
+      "step": 14621
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014290161892901619,
+      "loss": 0.2485,
+      "step": 14622
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00014288916562889164,
+      "loss": 0.189,
+      "step": 14623
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014287671232876713,
+      "loss": 0.188,
+      "step": 14624
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0001428642590286426,
+      "loss": 0.1198,
+      "step": 14625
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014285180572851805,
+      "loss": 0.1816,
+      "step": 14626
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0001428393524283935,
+      "loss": 0.3118,
+      "step": 14627
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014282689912826897,
+      "loss": 0.1805,
+      "step": 14628
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014281444582814445,
+      "loss": 0.2267,
+      "step": 14629
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0001428019925280199,
+      "loss": 0.2037,
+      "step": 14630
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014278953922789537,
+      "loss": 0.1794,
+      "step": 14631
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014277708592777086,
+      "loss": 0.2417,
+      "step": 14632
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014276463262764632,
+      "loss": 0.3166,
+      "step": 14633
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014275217932752178,
+      "loss": 0.2911,
+      "step": 14634
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014273972602739726,
+      "loss": 0.1678,
+      "step": 14635
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014272727272727272,
+      "loss": 0.2308,
+      "step": 14636
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014271481942714818,
+      "loss": 0.2056,
+      "step": 14637
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014270236612702364,
+      "loss": 0.0909,
+      "step": 14638
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014268991282689913,
+      "loss": 0.2623,
+      "step": 14639
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014267745952677459,
+      "loss": 0.3044,
+      "step": 14640
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014266500622665005,
+      "loss": 0.2359,
+      "step": 14641
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0001426525529265255,
+      "loss": 0.1487,
+      "step": 14642
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.000142640099626401,
+      "loss": 0.1709,
+      "step": 14643
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014262764632627645,
+      "loss": 0.1583,
+      "step": 14644
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0001426151930261519,
+      "loss": 0.3917,
+      "step": 14645
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0001426027397260274,
+      "loss": 0.2505,
+      "step": 14646
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014259028642590285,
+      "loss": 0.1835,
+      "step": 14647
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0001425778331257783,
+      "loss": 0.198,
+      "step": 14648
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00014256537982565377,
+      "loss": 0.179,
+      "step": 14649
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014255292652552926,
+      "loss": 0.2194,
+      "step": 14650
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014254047322540472,
+      "loss": 0.2035,
+      "step": 14651
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014252801992528018,
+      "loss": 0.2485,
+      "step": 14652
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014251556662515566,
+      "loss": 0.1752,
+      "step": 14653
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014250311332503112,
+      "loss": 0.1851,
+      "step": 14654
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014249066002490658,
+      "loss": 0.1227,
+      "step": 14655
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014247820672478207,
+      "loss": 0.163,
+      "step": 14656
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014246575342465753,
+      "loss": 0.1988,
+      "step": 14657
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.000142453300124533,
+      "loss": 0.1731,
+      "step": 14658
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014244084682440845,
+      "loss": 0.1935,
+      "step": 14659
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0001424283935242839,
+      "loss": 0.1414,
+      "step": 14660
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0001424159402241594,
+      "loss": 0.271,
+      "step": 14661
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014240348692403485,
+      "loss": 0.1516,
+      "step": 14662
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0001423910336239103,
+      "loss": 0.1561,
+      "step": 14663
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0001423785803237858,
+      "loss": 0.2501,
+      "step": 14664
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014236612702366125,
+      "loss": 0.1784,
+      "step": 14665
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014235367372353671,
+      "loss": 0.2042,
+      "step": 14666
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0001423412204234122,
+      "loss": 0.1861,
+      "step": 14667
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014232876712328766,
+      "loss": 0.1756,
+      "step": 14668
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014231631382316312,
+      "loss": 0.1193,
+      "step": 14669
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0001423038605230386,
+      "loss": 0.3953,
+      "step": 14670
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014229140722291406,
+      "loss": 0.243,
+      "step": 14671
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014227895392278952,
+      "loss": 0.2993,
+      "step": 14672
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014226650062266498,
+      "loss": 0.1897,
+      "step": 14673
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014225404732254044,
+      "loss": 0.3182,
+      "step": 14674
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00014224159402241593,
+      "loss": 0.2367,
+      "step": 14675
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0001422291407222914,
+      "loss": 0.2163,
+      "step": 14676
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014221668742216685,
+      "loss": 0.3326,
+      "step": 14677
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014220423412204233,
+      "loss": 0.1604,
+      "step": 14678
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0001421917808219178,
+      "loss": 0.1779,
+      "step": 14679
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014217932752179325,
+      "loss": 0.1535,
+      "step": 14680
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014216687422166874,
+      "loss": 0.149,
+      "step": 14681
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0001421544209215442,
+      "loss": 0.1065,
+      "step": 14682
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014214196762141966,
+      "loss": 0.3497,
+      "step": 14683
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014212951432129514,
+      "loss": 0.191,
+      "step": 14684
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0001421170610211706,
+      "loss": 0.1725,
+      "step": 14685
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014210460772104606,
+      "loss": 0.1259,
+      "step": 14686
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014209215442092155,
+      "loss": 0.238,
+      "step": 14687
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.000142079701120797,
+      "loss": 0.1276,
+      "step": 14688
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014206724782067246,
+      "loss": 0.2458,
+      "step": 14689
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014205479452054792,
+      "loss": 0.2116,
+      "step": 14690
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014204234122042338,
+      "loss": 0.261,
+      "step": 14691
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014202988792029887,
+      "loss": 0.2173,
+      "step": 14692
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014201743462017433,
+      "loss": 0.1443,
+      "step": 14693
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0001420049813200498,
+      "loss": 0.158,
+      "step": 14694
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014199252801992527,
+      "loss": 0.1771,
+      "step": 14695
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014198007471980073,
+      "loss": 0.3061,
+      "step": 14696
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0001419676214196762,
+      "loss": 0.1473,
+      "step": 14697
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014195516811955168,
+      "loss": 0.1872,
+      "step": 14698
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014194271481942714,
+      "loss": 0.2145,
+      "step": 14699
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0001419302615193026,
+      "loss": 0.1464,
+      "step": 14700
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00014191780821917808,
+      "loss": 0.2286,
+      "step": 14701
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014190535491905354,
+      "loss": 0.244,
+      "step": 14702
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.000141892901618929,
+      "loss": 0.1769,
+      "step": 14703
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014188044831880446,
+      "loss": 0.1943,
+      "step": 14704
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014186799501867992,
+      "loss": 0.1965,
+      "step": 14705
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0001418555417185554,
+      "loss": 0.1875,
+      "step": 14706
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014184308841843087,
+      "loss": 0.168,
+      "step": 14707
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014183063511830632,
+      "loss": 0.3977,
+      "step": 14708
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0001418181818181818,
+      "loss": 0.2834,
+      "step": 14709
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014180572851805727,
+      "loss": 0.1735,
+      "step": 14710
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014179327521793273,
+      "loss": 0.1543,
+      "step": 14711
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014178082191780822,
+      "loss": 0.1703,
+      "step": 14712
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014176836861768367,
+      "loss": 0.2542,
+      "step": 14713
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014175591531755913,
+      "loss": 0.4204,
+      "step": 14714
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014174346201743462,
+      "loss": 0.1876,
+      "step": 14715
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014173100871731008,
+      "loss": 0.2756,
+      "step": 14716
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014171855541718554,
+      "loss": 0.2025,
+      "step": 14717
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014170610211706102,
+      "loss": 0.1258,
+      "step": 14718
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014169364881693648,
+      "loss": 0.2417,
+      "step": 14719
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014168119551681194,
+      "loss": 0.2866,
+      "step": 14720
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0001416687422166874,
+      "loss": 0.2122,
+      "step": 14721
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014165628891656286,
+      "loss": 0.1666,
+      "step": 14722
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014164383561643835,
+      "loss": 0.1447,
+      "step": 14723
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0001416313823163138,
+      "loss": 0.1502,
+      "step": 14724
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014161892901618927,
+      "loss": 0.1553,
+      "step": 14725
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00014160647571606475,
+      "loss": 0.1621,
+      "step": 14726
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0001415940224159402,
+      "loss": 0.2238,
+      "step": 14727
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014158156911581567,
+      "loss": 0.194,
+      "step": 14728
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014156911581569116,
+      "loss": 0.1566,
+      "step": 14729
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014155666251556662,
+      "loss": 0.121,
+      "step": 14730
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014154420921544208,
+      "loss": 0.2052,
+      "step": 14731
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014153175591531756,
+      "loss": 0.1952,
+      "step": 14732
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014151930261519302,
+      "loss": 0.4085,
+      "step": 14733
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014150684931506848,
+      "loss": 0.146,
+      "step": 14734
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014149439601494397,
+      "loss": 0.1628,
+      "step": 14735
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0001414819427148194,
+      "loss": 0.1483,
+      "step": 14736
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014146948941469488,
+      "loss": 0.1654,
+      "step": 14737
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014145703611457034,
+      "loss": 0.2189,
+      "step": 14738
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0001414445828144458,
+      "loss": 0.2572,
+      "step": 14739
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0001414321295143213,
+      "loss": 0.1612,
+      "step": 14740
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014141967621419675,
+      "loss": 0.1557,
+      "step": 14741
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0001414072229140722,
+      "loss": 0.1497,
+      "step": 14742
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0001413947696139477,
+      "loss": 0.1761,
+      "step": 14743
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014138231631382315,
+      "loss": 0.1298,
+      "step": 14744
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0001413698630136986,
+      "loss": 0.2718,
+      "step": 14745
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0001413574097135741,
+      "loss": 0.1684,
+      "step": 14746
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014134495641344956,
+      "loss": 0.2521,
+      "step": 14747
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014133250311332502,
+      "loss": 0.1497,
+      "step": 14748
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0001413200498132005,
+      "loss": 0.1402,
+      "step": 14749
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014130759651307596,
+      "loss": 0.1365,
+      "step": 14750
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014129514321295142,
+      "loss": 0.1752,
+      "step": 14751
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014128268991282688,
+      "loss": 0.3409,
+      "step": 14752
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.00014127023661270234,
+      "loss": 0.1899,
+      "step": 14753
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014125778331257783,
+      "loss": 0.2393,
+      "step": 14754
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014124533001245328,
+      "loss": 0.1413,
+      "step": 14755
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014123287671232874,
+      "loss": 0.1705,
+      "step": 14756
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014122042341220423,
+      "loss": 0.1434,
+      "step": 14757
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0001412079701120797,
+      "loss": 0.286,
+      "step": 14758
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014119551681195515,
+      "loss": 0.1823,
+      "step": 14759
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014118306351183063,
+      "loss": 0.1559,
+      "step": 14760
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0001411706102117061,
+      "loss": 0.1273,
+      "step": 14761
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014115815691158155,
+      "loss": 0.1474,
+      "step": 14762
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014114570361145704,
+      "loss": 0.1313,
+      "step": 14763
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0001411332503113325,
+      "loss": 0.2758,
+      "step": 14764
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014112079701120796,
+      "loss": 0.2606,
+      "step": 14765
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014110834371108344,
+      "loss": 0.1997,
+      "step": 14766
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014109589041095888,
+      "loss": 0.1782,
+      "step": 14767
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014108343711083436,
+      "loss": 0.1597,
+      "step": 14768
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014107098381070982,
+      "loss": 0.1699,
+      "step": 14769
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014105853051058528,
+      "loss": 0.1808,
+      "step": 14770
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014104607721046077,
+      "loss": 0.1742,
+      "step": 14771
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014103362391033623,
+      "loss": 0.2793,
+      "step": 14772
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014102117061021169,
+      "loss": 0.1982,
+      "step": 14773
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014100871731008717,
+      "loss": 0.1509,
+      "step": 14774
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014099626400996263,
+      "loss": 0.152,
+      "step": 14775
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0001409838107098381,
+      "loss": 0.1618,
+      "step": 14776
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014097135740971358,
+      "loss": 0.2775,
+      "step": 14777
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.00014095890410958904,
+      "loss": 0.1263,
+      "step": 14778
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0001409464508094645,
+      "loss": 0.2828,
+      "step": 14779
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014093399750933998,
+      "loss": 0.1936,
+      "step": 14780
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014092154420921544,
+      "loss": 0.1563,
+      "step": 14781
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0001409090909090909,
+      "loss": 0.1764,
+      "step": 14782
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014089663760896636,
+      "loss": 0.3383,
+      "step": 14783
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014088418430884182,
+      "loss": 0.1913,
+      "step": 14784
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0001408717310087173,
+      "loss": 0.1715,
+      "step": 14785
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014085927770859276,
+      "loss": 0.1475,
+      "step": 14786
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014084682440846822,
+      "loss": 0.1493,
+      "step": 14787
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0001408343711083437,
+      "loss": 0.1666,
+      "step": 14788
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014082191780821917,
+      "loss": 0.3637,
+      "step": 14789
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014080946450809463,
+      "loss": 0.1462,
+      "step": 14790
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0001407970112079701,
+      "loss": 0.168,
+      "step": 14791
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014078455790784557,
+      "loss": 0.283,
+      "step": 14792
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014077210460772103,
+      "loss": 0.128,
+      "step": 14793
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014075965130759652,
+      "loss": 0.1939,
+      "step": 14794
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014074719800747198,
+      "loss": 0.2963,
+      "step": 14795
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014073474470734744,
+      "loss": 0.2425,
+      "step": 14796
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014072229140722292,
+      "loss": 0.1751,
+      "step": 14797
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014070983810709835,
+      "loss": 0.1412,
+      "step": 14798
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014069738480697384,
+      "loss": 0.2206,
+      "step": 14799
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0001406849315068493,
+      "loss": 0.1965,
+      "step": 14800
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014067247820672476,
+      "loss": 0.2826,
+      "step": 14801
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014066002490660025,
+      "loss": 0.2958,
+      "step": 14802
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0001406475716064757,
+      "loss": 0.1858,
+      "step": 14803
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014063511830635116,
+      "loss": 0.2135,
+      "step": 14804
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00014062266500622665,
+      "loss": 0.2495,
+      "step": 14805
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0001406102117061021,
+      "loss": 0.0943,
+      "step": 14806
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014059775840597757,
+      "loss": 0.1706,
+      "step": 14807
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014058530510585305,
+      "loss": 0.2734,
+      "step": 14808
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014057285180572851,
+      "loss": 0.2634,
+      "step": 14809
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014056039850560397,
+      "loss": 0.1756,
+      "step": 14810
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014054794520547943,
+      "loss": 0.1605,
+      "step": 14811
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014053549190535492,
+      "loss": 0.1493,
+      "step": 14812
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014052303860523038,
+      "loss": 0.1536,
+      "step": 14813
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014051058530510584,
+      "loss": 0.2754,
+      "step": 14814
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0001404981320049813,
+      "loss": 0.1404,
+      "step": 14815
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014048567870485678,
+      "loss": 0.1617,
+      "step": 14816
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014047322540473224,
+      "loss": 0.2726,
+      "step": 14817
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0001404607721046077,
+      "loss": 0.2231,
+      "step": 14818
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0001404483188044832,
+      "loss": 0.425,
+      "step": 14819
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014043586550435865,
+      "loss": 0.3502,
+      "step": 14820
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0001404234122042341,
+      "loss": 0.2718,
+      "step": 14821
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014041095890410956,
+      "loss": 0.1516,
+      "step": 14822
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014039850560398505,
+      "loss": 0.1816,
+      "step": 14823
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0001403860523038605,
+      "loss": 0.1451,
+      "step": 14824
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014037359900373597,
+      "loss": 0.148,
+      "step": 14825
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014036114570361146,
+      "loss": 0.1911,
+      "step": 14826
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014034869240348691,
+      "loss": 0.3477,
+      "step": 14827
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014033623910336237,
+      "loss": 0.1764,
+      "step": 14828
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014032378580323786,
+      "loss": 0.1923,
+      "step": 14829
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0001403113325031133,
+      "loss": 0.1768,
+      "step": 14830
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014029887920298878,
+      "loss": 0.1793,
+      "step": 14831
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00014028642590286424,
+      "loss": 0.2251,
+      "step": 14832
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0001402739726027397,
+      "loss": 0.3262,
+      "step": 14833
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014026151930261518,
+      "loss": 0.2116,
+      "step": 14834
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014024906600249064,
+      "loss": 0.2367,
+      "step": 14835
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0001402366127023661,
+      "loss": 0.2095,
+      "step": 14836
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0001402241594022416,
+      "loss": 0.2062,
+      "step": 14837
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014021170610211705,
+      "loss": 0.1843,
+      "step": 14838
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0001401992528019925,
+      "loss": 0.2294,
+      "step": 14839
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.000140186799501868,
+      "loss": 0.2219,
+      "step": 14840
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014017434620174345,
+      "loss": 0.2127,
+      "step": 14841
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0001401618929016189,
+      "loss": 0.2494,
+      "step": 14842
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0001401494396014944,
+      "loss": 0.1871,
+      "step": 14843
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014013698630136986,
+      "loss": 0.116,
+      "step": 14844
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014012453300124531,
+      "loss": 0.2005,
+      "step": 14845
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014011207970112077,
+      "loss": 0.187,
+      "step": 14846
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014009962640099623,
+      "loss": 0.2271,
+      "step": 14847
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014008717310087172,
+      "loss": 0.1716,
+      "step": 14848
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014007471980074718,
+      "loss": 0.1436,
+      "step": 14849
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014006226650062264,
+      "loss": 0.1016,
+      "step": 14850
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014004981320049812,
+      "loss": 0.1838,
+      "step": 14851
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014003735990037358,
+      "loss": 0.2579,
+      "step": 14852
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014002490660024904,
+      "loss": 0.1604,
+      "step": 14853
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014001245330012453,
+      "loss": 0.1325,
+      "step": 14854
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00014,
+      "loss": 0.1704,
+      "step": 14855
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00013998754669987545,
+      "loss": 0.1502,
+      "step": 14856
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00013997509339975093,
+      "loss": 0.2835,
+      "step": 14857
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0001399626400996264,
+      "loss": 0.3506,
+      "step": 14858
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013995018679950185,
+      "loss": 0.1761,
+      "step": 14859
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013993773349937734,
+      "loss": 0.1709,
+      "step": 14860
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013992528019925277,
+      "loss": 0.1449,
+      "step": 14861
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013991282689912826,
+      "loss": 0.1942,
+      "step": 14862
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013990037359900372,
+      "loss": 0.1167,
+      "step": 14863
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013988792029887917,
+      "loss": 0.2899,
+      "step": 14864
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013987546699875466,
+      "loss": 0.1645,
+      "step": 14865
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013986301369863012,
+      "loss": 0.1503,
+      "step": 14866
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013985056039850558,
+      "loss": 0.2716,
+      "step": 14867
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013983810709838107,
+      "loss": 0.2929,
+      "step": 14868
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013982565379825652,
+      "loss": 0.1596,
+      "step": 14869
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013981320049813198,
+      "loss": 0.2968,
+      "step": 14870
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013980074719800747,
+      "loss": 0.3116,
+      "step": 14871
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013978829389788293,
+      "loss": 0.3017,
+      "step": 14872
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0001397758405977584,
+      "loss": 0.1755,
+      "step": 14873
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013976338729763387,
+      "loss": 0.2549,
+      "step": 14874
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013975093399750933,
+      "loss": 0.1298,
+      "step": 14875
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0001397384806973848,
+      "loss": 0.2766,
+      "step": 14876
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013972602739726025,
+      "loss": 0.2021,
+      "step": 14877
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0001397135740971357,
+      "loss": 0.1555,
+      "step": 14878
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0001397011207970112,
+      "loss": 0.1647,
+      "step": 14879
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013968866749688666,
+      "loss": 0.1348,
+      "step": 14880
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013967621419676212,
+      "loss": 0.166,
+      "step": 14881
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0001396637608966376,
+      "loss": 0.1477,
+      "step": 14882
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013965130759651306,
+      "loss": 0.2529,
+      "step": 14883
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00013963885429638852,
+      "loss": 0.2125,
+      "step": 14884
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.000139626400996264,
+      "loss": 0.1972,
+      "step": 14885
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013961394769613947,
+      "loss": 0.1822,
+      "step": 14886
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013960149439601493,
+      "loss": 0.1784,
+      "step": 14887
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0001395890410958904,
+      "loss": 0.1743,
+      "step": 14888
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013957658779576587,
+      "loss": 0.3312,
+      "step": 14889
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013956413449564133,
+      "loss": 0.2211,
+      "step": 14890
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013955168119551682,
+      "loss": 0.1615,
+      "step": 14891
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013953922789539228,
+      "loss": 0.1622,
+      "step": 14892
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013952677459526773,
+      "loss": 0.2892,
+      "step": 14893
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0001395143212951432,
+      "loss": 0.214,
+      "step": 14894
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013950186799501865,
+      "loss": 0.3608,
+      "step": 14895
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013948941469489414,
+      "loss": 0.2233,
+      "step": 14896
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0001394769613947696,
+      "loss": 0.179,
+      "step": 14897
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013946450809464506,
+      "loss": 0.2386,
+      "step": 14898
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013945205479452054,
+      "loss": 0.2016,
+      "step": 14899
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.000139439601494396,
+      "loss": 0.1729,
+      "step": 14900
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013942714819427146,
+      "loss": 0.2831,
+      "step": 14901
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013941469489414695,
+      "loss": 0.3613,
+      "step": 14902
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0001394022415940224,
+      "loss": 0.167,
+      "step": 14903
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013938978829389787,
+      "loss": 0.1973,
+      "step": 14904
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013937733499377335,
+      "loss": 0.1443,
+      "step": 14905
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0001393648816936488,
+      "loss": 0.1405,
+      "step": 14906
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013935242839352427,
+      "loss": 0.1425,
+      "step": 14907
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013933997509339973,
+      "loss": 0.2847,
+      "step": 14908
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0001393275217932752,
+      "loss": 0.2732,
+      "step": 14909
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00013931506849315068,
+      "loss": 0.2365,
+      "step": 14910
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013930261519302613,
+      "loss": 0.1085,
+      "step": 14911
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0001392901618929016,
+      "loss": 0.3313,
+      "step": 14912
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013927770859277708,
+      "loss": 0.1936,
+      "step": 14913
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013926525529265254,
+      "loss": 0.3722,
+      "step": 14914
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.000139252801992528,
+      "loss": 0.2648,
+      "step": 14915
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013924034869240348,
+      "loss": 0.1467,
+      "step": 14916
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013922789539227894,
+      "loss": 0.1775,
+      "step": 14917
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0001392154420921544,
+      "loss": 0.1344,
+      "step": 14918
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0001392029887920299,
+      "loss": 0.1695,
+      "step": 14919
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013919053549190535,
+      "loss": 0.21,
+      "step": 14920
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0001391780821917808,
+      "loss": 0.2351,
+      "step": 14921
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0001391656288916563,
+      "loss": 0.2058,
+      "step": 14922
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013915317559153175,
+      "loss": 0.1575,
+      "step": 14923
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0001391407222914072,
+      "loss": 0.1516,
+      "step": 14924
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013912826899128267,
+      "loss": 0.1337,
+      "step": 14925
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013911581569115813,
+      "loss": 0.1716,
+      "step": 14926
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013910336239103362,
+      "loss": 0.2631,
+      "step": 14927
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013909090909090908,
+      "loss": 0.2389,
+      "step": 14928
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013907845579078454,
+      "loss": 0.2166,
+      "step": 14929
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013906600249066002,
+      "loss": 0.2332,
+      "step": 14930
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013905354919053548,
+      "loss": 0.1485,
+      "step": 14931
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013904109589041094,
+      "loss": 0.2147,
+      "step": 14932
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013902864259028643,
+      "loss": 0.2242,
+      "step": 14933
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013901618929016189,
+      "loss": 0.277,
+      "step": 14934
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013900373599003734,
+      "loss": 0.1949,
+      "step": 14935
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00013899128268991283,
+      "loss": 0.1567,
+      "step": 14936
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0001389788293897883,
+      "loss": 0.1773,
+      "step": 14937
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013896637608966375,
+      "loss": 0.0926,
+      "step": 14938
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013895392278953924,
+      "loss": 0.2601,
+      "step": 14939
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013894146948941467,
+      "loss": 0.1841,
+      "step": 14940
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013892901618929015,
+      "loss": 0.15,
+      "step": 14941
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0001389165628891656,
+      "loss": 0.1662,
+      "step": 14942
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013890410958904107,
+      "loss": 0.1833,
+      "step": 14943
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013889165628891656,
+      "loss": 0.2878,
+      "step": 14944
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013887920298879202,
+      "loss": 0.3357,
+      "step": 14945
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013886674968866748,
+      "loss": 0.2814,
+      "step": 14946
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013885429638854296,
+      "loss": 0.181,
+      "step": 14947
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013884184308841842,
+      "loss": 0.1986,
+      "step": 14948
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013882938978829388,
+      "loss": 0.1281,
+      "step": 14949
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013881693648816937,
+      "loss": 0.227,
+      "step": 14950
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013880448318804483,
+      "loss": 0.2349,
+      "step": 14951
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013879202988792029,
+      "loss": 0.345,
+      "step": 14952
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013877957658779577,
+      "loss": 0.232,
+      "step": 14953
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013876712328767123,
+      "loss": 0.1241,
+      "step": 14954
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0001387546699875467,
+      "loss": 0.2621,
+      "step": 14955
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013874221668742215,
+      "loss": 0.2465,
+      "step": 14956
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0001387297633872976,
+      "loss": 0.2006,
+      "step": 14957
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0001387173100871731,
+      "loss": 0.4075,
+      "step": 14958
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013870485678704855,
+      "loss": 0.1717,
+      "step": 14959
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013869240348692401,
+      "loss": 0.1931,
+      "step": 14960
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0001386799501867995,
+      "loss": 0.2061,
+      "step": 14961
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.00013866749688667496,
+      "loss": 0.1513,
+      "step": 14962
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013865504358655042,
+      "loss": 0.1927,
+      "step": 14963
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0001386425902864259,
+      "loss": 0.3497,
+      "step": 14964
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013863013698630136,
+      "loss": 0.2207,
+      "step": 14965
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013861768368617682,
+      "loss": 0.1431,
+      "step": 14966
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0001386052303860523,
+      "loss": 0.1155,
+      "step": 14967
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013859277708592777,
+      "loss": 0.1649,
+      "step": 14968
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013858032378580323,
+      "loss": 0.238,
+      "step": 14969
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013856787048567871,
+      "loss": 0.2905,
+      "step": 14970
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013855541718555415,
+      "loss": 0.1981,
+      "step": 14971
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013854296388542963,
+      "loss": 0.1132,
+      "step": 14972
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0001385305105853051,
+      "loss": 0.186,
+      "step": 14973
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013851805728518055,
+      "loss": 0.2309,
+      "step": 14974
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013850560398505604,
+      "loss": 0.1582,
+      "step": 14975
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0001384931506849315,
+      "loss": 0.2045,
+      "step": 14976
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013848069738480695,
+      "loss": 0.284,
+      "step": 14977
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013846824408468244,
+      "loss": 0.1713,
+      "step": 14978
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0001384557907845579,
+      "loss": 0.171,
+      "step": 14979
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013844333748443336,
+      "loss": 0.2291,
+      "step": 14980
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013843088418430882,
+      "loss": 0.1368,
+      "step": 14981
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0001384184308841843,
+      "loss": 0.1885,
+      "step": 14982
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013840597758405976,
+      "loss": 0.2696,
+      "step": 14983
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013839352428393522,
+      "loss": 0.1918,
+      "step": 14984
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0001383810709838107,
+      "loss": 0.2134,
+      "step": 14985
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013836861768368617,
+      "loss": 0.154,
+      "step": 14986
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00013835616438356163,
+      "loss": 0.2898,
+      "step": 14987
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0001383437110834371,
+      "loss": 0.1531,
+      "step": 14988
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013833125778331257,
+      "loss": 0.2803,
+      "step": 14989
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013831880448318803,
+      "loss": 0.1593,
+      "step": 14990
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0001383063511830635,
+      "loss": 0.1587,
+      "step": 14991
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013829389788293895,
+      "loss": 0.1416,
+      "step": 14992
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013828144458281444,
+      "loss": 0.1761,
+      "step": 14993
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0001382689912826899,
+      "loss": 0.1681,
+      "step": 14994
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013825653798256536,
+      "loss": 0.2212,
+      "step": 14995
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013824408468244084,
+      "loss": 0.2444,
+      "step": 14996
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0001382316313823163,
+      "loss": 0.1884,
+      "step": 14997
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013821917808219176,
+      "loss": 0.1356,
+      "step": 14998
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013820672478206725,
+      "loss": 0.135,
+      "step": 14999
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0001381942714819427,
+      "loss": 0.1898,
+      "step": 15000
+    },
+    {
+      "epoch": 5.75,
+      "eval_cer": 0.10495346976998264,
+      "eval_loss": 0.3659653663635254,
+      "eval_runtime": 47.1376,
+      "eval_samples_per_second": 37.189,
+      "eval_steps_per_second": 4.667,
+      "eval_wer": 0.1725573607277006,
+      "step": 15000
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013818181818181816,
+      "loss": 0.1542,
+      "step": 15001
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013816936488169365,
+      "loss": 0.375,
+      "step": 15002
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013815691158156908,
+      "loss": 0.2244,
+      "step": 15003
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013814445828144457,
+      "loss": 0.176,
+      "step": 15004
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013813200498132003,
+      "loss": 0.1804,
+      "step": 15005
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0001381195516811955,
+      "loss": 0.1275,
+      "step": 15006
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013810709838107097,
+      "loss": 0.2311,
+      "step": 15007
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013809464508094643,
+      "loss": 0.3019,
+      "step": 15008
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0001380821917808219,
+      "loss": 0.1826,
+      "step": 15009
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013806973848069738,
+      "loss": 0.2613,
+      "step": 15010
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013805728518057284,
+      "loss": 0.2383,
+      "step": 15011
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0001380448318804483,
+      "loss": 0.1212,
+      "step": 15012
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013803237858032378,
+      "loss": 0.1753,
+      "step": 15013
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00013801992528019924,
+      "loss": 0.2606,
+      "step": 15014
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0001380074719800747,
+      "loss": 0.1812,
+      "step": 15015
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0001379950186799502,
+      "loss": 0.2369,
+      "step": 15016
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013798256537982565,
+      "loss": 0.2359,
+      "step": 15017
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0001379701120797011,
+      "loss": 0.1899,
+      "step": 15018
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013795765877957657,
+      "loss": 0.1649,
+      "step": 15019
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013794520547945202,
+      "loss": 0.2452,
+      "step": 15020
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0001379327521793275,
+      "loss": 0.178,
+      "step": 15021
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013792029887920297,
+      "loss": 0.1403,
+      "step": 15022
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013790784557907843,
+      "loss": 0.2087,
+      "step": 15023
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013789539227895392,
+      "loss": 0.231,
+      "step": 15024
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013788293897882937,
+      "loss": 0.1542,
+      "step": 15025
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013787048567870483,
+      "loss": 0.2671,
+      "step": 15026
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013785803237858032,
+      "loss": 0.6102,
+      "step": 15027
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013784557907845578,
+      "loss": 0.143,
+      "step": 15028
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013783312577833124,
+      "loss": 0.1978,
+      "step": 15029
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013782067247820672,
+      "loss": 0.3106,
+      "step": 15030
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013780821917808218,
+      "loss": 0.1469,
+      "step": 15031
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013779576587795764,
+      "loss": 0.1917,
+      "step": 15032
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013778331257783313,
+      "loss": 0.3488,
+      "step": 15033
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013777085927770856,
+      "loss": 0.1544,
+      "step": 15034
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013775840597758405,
+      "loss": 0.1801,
+      "step": 15035
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0001377459526774595,
+      "loss": 0.1483,
+      "step": 15036
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013773349937733497,
+      "loss": 0.1786,
+      "step": 15037
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013772104607721045,
+      "loss": 0.1938,
+      "step": 15038
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0001377085927770859,
+      "loss": 0.2895,
+      "step": 15039
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00013769613947696137,
+      "loss": 0.2023,
+      "step": 15040
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013768368617683686,
+      "loss": 0.1881,
+      "step": 15041
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013767123287671232,
+      "loss": 0.1625,
+      "step": 15042
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013765877957658778,
+      "loss": 0.1945,
+      "step": 15043
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013764632627646326,
+      "loss": 0.1345,
+      "step": 15044
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013763387297633872,
+      "loss": 0.2648,
+      "step": 15045
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013762141967621418,
+      "loss": 0.2027,
+      "step": 15046
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013760896637608967,
+      "loss": 0.1776,
+      "step": 15047
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013759651307596513,
+      "loss": 0.1612,
+      "step": 15048
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013758405977584058,
+      "loss": 0.2451,
+      "step": 15049
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013757160647571604,
+      "loss": 0.1522,
+      "step": 15050
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0001375591531755915,
+      "loss": 0.3288,
+      "step": 15051
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.000137546699875467,
+      "loss": 0.1679,
+      "step": 15052
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013753424657534245,
+      "loss": 0.1887,
+      "step": 15053
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0001375217932752179,
+      "loss": 0.1765,
+      "step": 15054
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0001375093399750934,
+      "loss": 0.1464,
+      "step": 15055
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013749688667496885,
+      "loss": 0.1852,
+      "step": 15056
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0001374844333748443,
+      "loss": 0.1397,
+      "step": 15057
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0001374719800747198,
+      "loss": 0.3413,
+      "step": 15058
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013745952677459526,
+      "loss": 0.1567,
+      "step": 15059
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013744707347447072,
+      "loss": 0.251,
+      "step": 15060
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0001374346201743462,
+      "loss": 0.1253,
+      "step": 15061
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013742216687422166,
+      "loss": 0.1795,
+      "step": 15062
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013740971357409712,
+      "loss": 0.1748,
+      "step": 15063
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0001373972602739726,
+      "loss": 0.2454,
+      "step": 15064
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013738480697384804,
+      "loss": 0.1797,
+      "step": 15065
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00013737235367372353,
+      "loss": 0.1789,
+      "step": 15066
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013735990037359898,
+      "loss": 0.1356,
+      "step": 15067
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013734744707347444,
+      "loss": 0.1013,
+      "step": 15068
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013733499377334993,
+      "loss": 0.17,
+      "step": 15069
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0001373225404732254,
+      "loss": 0.2794,
+      "step": 15070
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013731008717310085,
+      "loss": 0.1721,
+      "step": 15071
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013729763387297633,
+      "loss": 0.1835,
+      "step": 15072
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0001372851805728518,
+      "loss": 0.1667,
+      "step": 15073
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013727272727272725,
+      "loss": 0.1504,
+      "step": 15074
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013726027397260274,
+      "loss": 0.1314,
+      "step": 15075
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0001372478206724782,
+      "loss": 0.1596,
+      "step": 15076
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013723536737235366,
+      "loss": 0.2617,
+      "step": 15077
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013722291407222914,
+      "loss": 0.2298,
+      "step": 15078
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0001372104607721046,
+      "loss": 0.2326,
+      "step": 15079
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013719800747198006,
+      "loss": 0.1234,
+      "step": 15080
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013718555417185552,
+      "loss": 0.1586,
+      "step": 15081
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013717310087173098,
+      "loss": 0.1815,
+      "step": 15082
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013716064757160647,
+      "loss": 0.3507,
+      "step": 15083
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013714819427148193,
+      "loss": 0.2063,
+      "step": 15084
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013713574097135739,
+      "loss": 0.1343,
+      "step": 15085
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013712328767123287,
+      "loss": 0.162,
+      "step": 15086
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013711083437110833,
+      "loss": 0.1355,
+      "step": 15087
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0001370983810709838,
+      "loss": 0.3334,
+      "step": 15088
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013708592777085928,
+      "loss": 0.3098,
+      "step": 15089
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013707347447073474,
+      "loss": 0.1697,
+      "step": 15090
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0001370610211706102,
+      "loss": 0.2157,
+      "step": 15091
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013704856787048568,
+      "loss": 0.1238,
+      "step": 15092
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00013703611457036114,
+      "loss": 0.1323,
+      "step": 15093
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0001370236612702366,
+      "loss": 0.1666,
+      "step": 15094
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013701120797011209,
+      "loss": 0.4454,
+      "step": 15095
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013699875466998754,
+      "loss": 0.172,
+      "step": 15096
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.000136986301369863,
+      "loss": 0.168,
+      "step": 15097
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013697384806973846,
+      "loss": 0.126,
+      "step": 15098
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013696139476961392,
+      "loss": 0.178,
+      "step": 15099
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0001369489414694894,
+      "loss": 0.2326,
+      "step": 15100
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013693648816936487,
+      "loss": 0.1645,
+      "step": 15101
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013692403486924033,
+      "loss": 0.2698,
+      "step": 15102
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0001369115815691158,
+      "loss": 0.1955,
+      "step": 15103
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013689912826899127,
+      "loss": 0.1693,
+      "step": 15104
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013688667496886673,
+      "loss": 0.1165,
+      "step": 15105
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013687422166874222,
+      "loss": 0.1591,
+      "step": 15106
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013686176836861768,
+      "loss": 0.1409,
+      "step": 15107
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013684931506849314,
+      "loss": 0.2784,
+      "step": 15108
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013683686176836862,
+      "loss": 0.1553,
+      "step": 15109
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013682440846824408,
+      "loss": 0.1743,
+      "step": 15110
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013681195516811954,
+      "loss": 0.136,
+      "step": 15111
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013679950186799503,
+      "loss": 0.2293,
+      "step": 15112
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013678704856787046,
+      "loss": 0.211,
+      "step": 15113
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013677459526774595,
+      "loss": 0.2124,
+      "step": 15114
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0001367621419676214,
+      "loss": 0.1765,
+      "step": 15115
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013674968866749686,
+      "loss": 0.2769,
+      "step": 15116
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013673723536737235,
+      "loss": 0.2351,
+      "step": 15117
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0001367247820672478,
+      "loss": 0.241,
+      "step": 15118
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00013671232876712327,
+      "loss": 0.1896,
+      "step": 15119
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013669987546699875,
+      "loss": 0.27,
+      "step": 15120
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013668742216687421,
+      "loss": 0.2987,
+      "step": 15121
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013667496886674967,
+      "loss": 0.2302,
+      "step": 15122
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013666251556662516,
+      "loss": 0.2267,
+      "step": 15123
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013665006226650062,
+      "loss": 0.2428,
+      "step": 15124
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013663760896637608,
+      "loss": 0.1354,
+      "step": 15125
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013662515566625156,
+      "loss": 0.2206,
+      "step": 15126
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013661270236612702,
+      "loss": 0.308,
+      "step": 15127
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013660024906600248,
+      "loss": 0.2218,
+      "step": 15128
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013658779576587794,
+      "loss": 0.2183,
+      "step": 15129
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0001365753424657534,
+      "loss": 0.1668,
+      "step": 15130
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0001365628891656289,
+      "loss": 0.2039,
+      "step": 15131
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013655043586550435,
+      "loss": 0.1933,
+      "step": 15132
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0001365379825653798,
+      "loss": 0.245,
+      "step": 15133
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0001365255292652553,
+      "loss": 0.1884,
+      "step": 15134
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013651307596513075,
+      "loss": 0.2042,
+      "step": 15135
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0001365006226650062,
+      "loss": 0.1923,
+      "step": 15136
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0001364881693648817,
+      "loss": 0.1408,
+      "step": 15137
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013647571606475716,
+      "loss": 0.2513,
+      "step": 15138
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013646326276463261,
+      "loss": 0.445,
+      "step": 15139
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0001364508094645081,
+      "loss": 0.2155,
+      "step": 15140
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013643835616438356,
+      "loss": 0.2244,
+      "step": 15141
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013642590286425902,
+      "loss": 0.1403,
+      "step": 15142
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013641344956413448,
+      "loss": 0.2034,
+      "step": 15143
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013640099626400994,
+      "loss": 0.1679,
+      "step": 15144
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.00013638854296388542,
+      "loss": 0.1685,
+      "step": 15145
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013637608966376088,
+      "loss": 0.2507,
+      "step": 15146
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013636363636363634,
+      "loss": 0.1962,
+      "step": 15147
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013635118306351183,
+      "loss": 0.2193,
+      "step": 15148
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0001363387297633873,
+      "loss": 0.1648,
+      "step": 15149
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013632627646326275,
+      "loss": 0.2015,
+      "step": 15150
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013631382316313823,
+      "loss": 0.1914,
+      "step": 15151
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0001363013698630137,
+      "loss": 0.3789,
+      "step": 15152
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013628891656288915,
+      "loss": 0.187,
+      "step": 15153
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0001362764632627646,
+      "loss": 0.2195,
+      "step": 15154
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0001362640099626401,
+      "loss": 0.1452,
+      "step": 15155
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013625155666251556,
+      "loss": 0.1247,
+      "step": 15156
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013623910336239101,
+      "loss": 0.2001,
+      "step": 15157
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0001362266500622665,
+      "loss": 0.3321,
+      "step": 15158
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013621419676214196,
+      "loss": 0.1519,
+      "step": 15159
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013620174346201742,
+      "loss": 0.3493,
+      "step": 15160
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013618929016189288,
+      "loss": 0.2501,
+      "step": 15161
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013617683686176836,
+      "loss": 0.1694,
+      "step": 15162
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013616438356164382,
+      "loss": 0.11,
+      "step": 15163
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013615193026151928,
+      "loss": 0.2983,
+      "step": 15164
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013613947696139474,
+      "loss": 0.2186,
+      "step": 15165
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013612702366127023,
+      "loss": 0.1903,
+      "step": 15166
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0001361145703611457,
+      "loss": 0.171,
+      "step": 15167
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013610211706102115,
+      "loss": 0.2793,
+      "step": 15168
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013608966376089663,
+      "loss": 0.1583,
+      "step": 15169
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0001360772104607721,
+      "loss": 0.4547,
+      "step": 15170
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00013606475716064755,
+      "loss": 0.2874,
+      "step": 15171
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013605230386052304,
+      "loss": 0.1945,
+      "step": 15172
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0001360398505603985,
+      "loss": 0.2101,
+      "step": 15173
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013602739726027396,
+      "loss": 0.1391,
+      "step": 15174
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013601494396014942,
+      "loss": 0.159,
+      "step": 15175
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013600249066002487,
+      "loss": 0.277,
+      "step": 15176
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013599003735990036,
+      "loss": 0.2587,
+      "step": 15177
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013597758405977582,
+      "loss": 0.1492,
+      "step": 15178
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013596513075965128,
+      "loss": 0.1571,
+      "step": 15179
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013595267745952677,
+      "loss": 0.3215,
+      "step": 15180
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013594022415940222,
+      "loss": 0.1988,
+      "step": 15181
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013592777085927768,
+      "loss": 0.3898,
+      "step": 15182
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013591531755915317,
+      "loss": 0.2685,
+      "step": 15183
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013590286425902863,
+      "loss": 0.2558,
+      "step": 15184
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0001358904109589041,
+      "loss": 0.167,
+      "step": 15185
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013587795765877957,
+      "loss": 0.1541,
+      "step": 15186
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013586550435865503,
+      "loss": 0.1469,
+      "step": 15187
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0001358530510585305,
+      "loss": 0.1334,
+      "step": 15188
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013584059775840598,
+      "loss": 0.2808,
+      "step": 15189
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013582814445828144,
+      "loss": 0.2333,
+      "step": 15190
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0001358156911581569,
+      "loss": 0.1597,
+      "step": 15191
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013580323785803236,
+      "loss": 0.1556,
+      "step": 15192
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013579078455790782,
+      "loss": 0.2385,
+      "step": 15193
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0001357783312577833,
+      "loss": 0.2451,
+      "step": 15194
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013576587795765876,
+      "loss": 0.1817,
+      "step": 15195
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00013575342465753422,
+      "loss": 0.1957,
+      "step": 15196
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0001357409713574097,
+      "loss": 0.3538,
+      "step": 15197
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013572851805728517,
+      "loss": 0.1659,
+      "step": 15198
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013571606475716063,
+      "loss": 0.2566,
+      "step": 15199
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0001357036114570361,
+      "loss": 0.2015,
+      "step": 15200
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013569115815691157,
+      "loss": 0.254,
+      "step": 15201
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013567870485678703,
+      "loss": 0.3523,
+      "step": 15202
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013566625155666252,
+      "loss": 0.2456,
+      "step": 15203
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013565379825653798,
+      "loss": 0.1778,
+      "step": 15204
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013564134495641343,
+      "loss": 0.2069,
+      "step": 15205
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013562889165628892,
+      "loss": 0.1645,
+      "step": 15206
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013561643835616435,
+      "loss": 0.3132,
+      "step": 15207
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013560398505603984,
+      "loss": 0.199,
+      "step": 15208
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0001355915317559153,
+      "loss": 0.1668,
+      "step": 15209
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013557907845579076,
+      "loss": 0.2156,
+      "step": 15210
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013556662515566624,
+      "loss": 0.1155,
+      "step": 15211
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0001355541718555417,
+      "loss": 0.1535,
+      "step": 15212
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013554171855541716,
+      "loss": 0.1587,
+      "step": 15213
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013552926525529265,
+      "loss": 0.2831,
+      "step": 15214
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0001355168119551681,
+      "loss": 0.2744,
+      "step": 15215
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013550435865504357,
+      "loss": 0.2458,
+      "step": 15216
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013549190535491905,
+      "loss": 0.169,
+      "step": 15217
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0001354794520547945,
+      "loss": 0.1506,
+      "step": 15218
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013546699875466997,
+      "loss": 0.1432,
+      "step": 15219
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013545454545454546,
+      "loss": 0.2917,
+      "step": 15220
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013544209215442092,
+      "loss": 0.1782,
+      "step": 15221
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013542963885429638,
+      "loss": 0.2525,
+      "step": 15222
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00013541718555417183,
+      "loss": 0.1888,
+      "step": 15223
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0001354047322540473,
+      "loss": 0.2124,
+      "step": 15224
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013539227895392278,
+      "loss": 0.1406,
+      "step": 15225
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013537982565379824,
+      "loss": 0.3374,
+      "step": 15226
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0001353673723536737,
+      "loss": 0.2521,
+      "step": 15227
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013535491905354919,
+      "loss": 0.2283,
+      "step": 15228
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013534246575342464,
+      "loss": 0.287,
+      "step": 15229
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0001353300124533001,
+      "loss": 0.1403,
+      "step": 15230
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0001353175591531756,
+      "loss": 0.1494,
+      "step": 15231
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013530510585305105,
+      "loss": 0.1397,
+      "step": 15232
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0001352926525529265,
+      "loss": 0.2349,
+      "step": 15233
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.000135280199252802,
+      "loss": 0.2558,
+      "step": 15234
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013526774595267745,
+      "loss": 0.1511,
+      "step": 15235
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0001352552926525529,
+      "loss": 0.112,
+      "step": 15236
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0001352428393524284,
+      "loss": 0.142,
+      "step": 15237
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013523038605230383,
+      "loss": 0.1802,
+      "step": 15238
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013521793275217932,
+      "loss": 0.2097,
+      "step": 15239
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013520547945205478,
+      "loss": 0.269,
+      "step": 15240
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013519302615193024,
+      "loss": 0.1966,
+      "step": 15241
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013518057285180572,
+      "loss": 0.2021,
+      "step": 15242
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013516811955168118,
+      "loss": 0.2349,
+      "step": 15243
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013515566625155664,
+      "loss": 0.1977,
+      "step": 15244
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013514321295143213,
+      "loss": 0.2648,
+      "step": 15245
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013513075965130759,
+      "loss": 0.2533,
+      "step": 15246
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013511830635118304,
+      "loss": 0.2643,
+      "step": 15247
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00013510585305105853,
+      "loss": 0.2152,
+      "step": 15248
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.000135093399750934,
+      "loss": 0.2411,
+      "step": 15249
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013508094645080945,
+      "loss": 0.1572,
+      "step": 15250
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013506849315068494,
+      "loss": 0.2422,
+      "step": 15251
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0001350560398505604,
+      "loss": 0.3167,
+      "step": 15252
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013504358655043585,
+      "loss": 0.1955,
+      "step": 15253
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0001350311332503113,
+      "loss": 0.1973,
+      "step": 15254
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013501867995018677,
+      "loss": 0.1576,
+      "step": 15255
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013500622665006226,
+      "loss": 0.1435,
+      "step": 15256
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013499377334993772,
+      "loss": 0.1866,
+      "step": 15257
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013498132004981318,
+      "loss": 0.3194,
+      "step": 15258
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013496886674968866,
+      "loss": 0.2653,
+      "step": 15259
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013495641344956412,
+      "loss": 0.1642,
+      "step": 15260
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013494396014943958,
+      "loss": 0.1754,
+      "step": 15261
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013493150684931507,
+      "loss": 0.1626,
+      "step": 15262
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013491905354919053,
+      "loss": 0.1271,
+      "step": 15263
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013490660024906599,
+      "loss": 0.2127,
+      "step": 15264
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013489414694894147,
+      "loss": 0.1652,
+      "step": 15265
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013488169364881693,
+      "loss": 0.2674,
+      "step": 15266
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0001348692403486924,
+      "loss": 0.1408,
+      "step": 15267
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013485678704856788,
+      "loss": 0.2041,
+      "step": 15268
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013484433374844334,
+      "loss": 0.1733,
+      "step": 15269
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0001348318804483188,
+      "loss": 0.303,
+      "step": 15270
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013481942714819425,
+      "loss": 0.2859,
+      "step": 15271
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013480697384806971,
+      "loss": 0.2773,
+      "step": 15272
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0001347945205479452,
+      "loss": 0.1588,
+      "step": 15273
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013478206724782066,
+      "loss": 0.2017,
+      "step": 15274
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00013476961394769612,
+      "loss": 0.2843,
+      "step": 15275
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0001347571606475716,
+      "loss": 0.2014,
+      "step": 15276
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013474470734744706,
+      "loss": 0.2345,
+      "step": 15277
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013473225404732252,
+      "loss": 0.2135,
+      "step": 15278
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.000134719800747198,
+      "loss": 0.1524,
+      "step": 15279
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013470734744707347,
+      "loss": 0.2133,
+      "step": 15280
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013469489414694893,
+      "loss": 0.2461,
+      "step": 15281
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013468244084682441,
+      "loss": 0.2539,
+      "step": 15282
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013466998754669987,
+      "loss": 0.2378,
+      "step": 15283
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013465753424657533,
+      "loss": 0.1938,
+      "step": 15284
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0001346450809464508,
+      "loss": 0.198,
+      "step": 15285
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013463262764632625,
+      "loss": 0.1394,
+      "step": 15286
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013462017434620174,
+      "loss": 0.2578,
+      "step": 15287
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0001346077210460772,
+      "loss": 0.17,
+      "step": 15288
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013459526774595266,
+      "loss": 0.2858,
+      "step": 15289
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013458281444582814,
+      "loss": 0.2357,
+      "step": 15290
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0001345703611457036,
+      "loss": 0.1802,
+      "step": 15291
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013455790784557906,
+      "loss": 0.2031,
+      "step": 15292
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013454545454545455,
+      "loss": 0.1408,
+      "step": 15293
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013453300124533,
+      "loss": 0.2146,
+      "step": 15294
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013452054794520546,
+      "loss": 0.3069,
+      "step": 15295
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013450809464508095,
+      "loss": 0.2422,
+      "step": 15296
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0001344956413449564,
+      "loss": 0.1926,
+      "step": 15297
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013448318804483187,
+      "loss": 0.1185,
+      "step": 15298
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013447073474470736,
+      "loss": 0.2131,
+      "step": 15299
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013445828144458281,
+      "loss": 0.1326,
+      "step": 15300
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.00013444582814445827,
+      "loss": 0.139,
+      "step": 15301
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013443337484433373,
+      "loss": 0.3498,
+      "step": 15302
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0001344209215442092,
+      "loss": 0.1802,
+      "step": 15303
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013440846824408468,
+      "loss": 0.2554,
+      "step": 15304
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013439601494396014,
+      "loss": 0.1036,
+      "step": 15305
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0001343835616438356,
+      "loss": 0.1868,
+      "step": 15306
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013437110834371108,
+      "loss": 0.1486,
+      "step": 15307
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013435865504358654,
+      "loss": 0.3227,
+      "step": 15308
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.000134346201743462,
+      "loss": 0.244,
+      "step": 15309
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0001343337484433375,
+      "loss": 0.2175,
+      "step": 15310
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013432129514321295,
+      "loss": 0.1521,
+      "step": 15311
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0001343088418430884,
+      "loss": 0.1845,
+      "step": 15312
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0001342963885429639,
+      "loss": 0.2486,
+      "step": 15313
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013428393524283935,
+      "loss": 0.2383,
+      "step": 15314
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0001342714819427148,
+      "loss": 0.1972,
+      "step": 15315
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013425902864259027,
+      "loss": 0.3803,
+      "step": 15316
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013424657534246573,
+      "loss": 0.1948,
+      "step": 15317
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013423412204234121,
+      "loss": 0.2264,
+      "step": 15318
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013422166874221667,
+      "loss": 0.1753,
+      "step": 15319
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013420921544209213,
+      "loss": 0.2758,
+      "step": 15320
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013419676214196762,
+      "loss": 0.1779,
+      "step": 15321
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013418430884184308,
+      "loss": 0.2075,
+      "step": 15322
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013417185554171854,
+      "loss": 0.1927,
+      "step": 15323
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013415940224159402,
+      "loss": 0.2229,
+      "step": 15324
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013414694894146948,
+      "loss": 0.2111,
+      "step": 15325
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00013413449564134494,
+      "loss": 0.1703,
+      "step": 15326
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0001341220423412204,
+      "loss": 0.2856,
+      "step": 15327
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0001341095890410959,
+      "loss": 0.2661,
+      "step": 15328
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013409713574097135,
+      "loss": 0.2312,
+      "step": 15329
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0001340846824408468,
+      "loss": 0.1547,
+      "step": 15330
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0001340722291407223,
+      "loss": 0.1217,
+      "step": 15331
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013405977584059775,
+      "loss": 0.1297,
+      "step": 15332
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0001340473225404732,
+      "loss": 0.3624,
+      "step": 15333
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013403486924034867,
+      "loss": 0.1473,
+      "step": 15334
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013402241594022413,
+      "loss": 0.17,
+      "step": 15335
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013400996264009962,
+      "loss": 0.1806,
+      "step": 15336
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013399750933997507,
+      "loss": 0.1681,
+      "step": 15337
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013398505603985053,
+      "loss": 0.2145,
+      "step": 15338
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013397260273972602,
+      "loss": 0.2182,
+      "step": 15339
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013396014943960148,
+      "loss": 0.1712,
+      "step": 15340
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013394769613947694,
+      "loss": 0.1825,
+      "step": 15341
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013393524283935242,
+      "loss": 0.1808,
+      "step": 15342
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013392278953922788,
+      "loss": 0.2473,
+      "step": 15343
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013391033623910334,
+      "loss": 0.2158,
+      "step": 15344
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013389788293897883,
+      "loss": 0.2897,
+      "step": 15345
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0001338854296388543,
+      "loss": 0.1615,
+      "step": 15346
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013387297633872975,
+      "loss": 0.1911,
+      "step": 15347
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0001338605230386052,
+      "loss": 0.1505,
+      "step": 15348
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013384806973848067,
+      "loss": 0.1405,
+      "step": 15349
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013383561643835615,
+      "loss": 0.1633,
+      "step": 15350
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0001338231631382316,
+      "loss": 0.1397,
+      "step": 15351
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013381070983810707,
+      "loss": 0.2923,
+      "step": 15352
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00013379825653798256,
+      "loss": 0.1502,
+      "step": 15353
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013378580323785802,
+      "loss": 0.1672,
+      "step": 15354
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013377334993773348,
+      "loss": 0.1541,
+      "step": 15355
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013376089663760896,
+      "loss": 0.1421,
+      "step": 15356
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013374844333748442,
+      "loss": 0.2033,
+      "step": 15357
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013373599003735988,
+      "loss": 0.2459,
+      "step": 15358
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013372353673723537,
+      "loss": 0.2953,
+      "step": 15359
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013371108343711083,
+      "loss": 0.1849,
+      "step": 15360
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013369863013698628,
+      "loss": 0.2183,
+      "step": 15361
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013368617683686177,
+      "loss": 0.1875,
+      "step": 15362
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013367372353673723,
+      "loss": 0.3484,
+      "step": 15363
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0001336612702366127,
+      "loss": 0.2807,
+      "step": 15364
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013364881693648815,
+      "loss": 0.1794,
+      "step": 15365
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0001336363636363636,
+      "loss": 0.2201,
+      "step": 15366
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0001336239103362391,
+      "loss": 0.1403,
+      "step": 15367
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013361145703611455,
+      "loss": 0.1626,
+      "step": 15368
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013359900373599,
+      "loss": 0.2291,
+      "step": 15369
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0001335865504358655,
+      "loss": 0.3015,
+      "step": 15370
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013357409713574096,
+      "loss": 0.2161,
+      "step": 15371
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013356164383561642,
+      "loss": 0.1814,
+      "step": 15372
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0001335491905354919,
+      "loss": 0.2515,
+      "step": 15373
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013353673723536736,
+      "loss": 0.1628,
+      "step": 15374
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013352428393524282,
+      "loss": 0.1798,
+      "step": 15375
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0001335118306351183,
+      "loss": 0.1841,
+      "step": 15376
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013349937733499377,
+      "loss": 0.3445,
+      "step": 15377
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013348692403486923,
+      "loss": 0.2071,
+      "step": 15378
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013347447073474469,
+      "loss": 0.258,
+      "step": 15379
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00013346201743462014,
+      "loss": 0.2284,
+      "step": 15380
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013344956413449563,
+      "loss": 0.2855,
+      "step": 15381
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0001334371108343711,
+      "loss": 0.2792,
+      "step": 15382
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013342465753424655,
+      "loss": 0.2813,
+      "step": 15383
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013341220423412204,
+      "loss": 0.1803,
+      "step": 15384
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0001333997509339975,
+      "loss": 0.1814,
+      "step": 15385
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013338729763387295,
+      "loss": 0.1767,
+      "step": 15386
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013337484433374844,
+      "loss": 0.1239,
+      "step": 15387
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0001333623910336239,
+      "loss": 0.2817,
+      "step": 15388
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013334993773349936,
+      "loss": 0.3536,
+      "step": 15389
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013333748443337484,
+      "loss": 0.1653,
+      "step": 15390
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0001333250311332503,
+      "loss": 0.1605,
+      "step": 15391
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013331257783312576,
+      "loss": 0.2126,
+      "step": 15392
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013330012453300125,
+      "loss": 0.1296,
+      "step": 15393
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0001332876712328767,
+      "loss": 0.1716,
+      "step": 15394
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013327521793275217,
+      "loss": 0.3744,
+      "step": 15395
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013326276463262763,
+      "loss": 0.2245,
+      "step": 15396
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013325031133250309,
+      "loss": 0.2307,
+      "step": 15397
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013323785803237857,
+      "loss": 0.207,
+      "step": 15398
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013322540473225403,
+      "loss": 0.197,
+      "step": 15399
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0001332129514321295,
+      "loss": 0.1475,
+      "step": 15400
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013320049813200498,
+      "loss": 0.2545,
+      "step": 15401
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013318804483188044,
+      "loss": 0.4148,
+      "step": 15402
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0001331755915317559,
+      "loss": 0.2671,
+      "step": 15403
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013316313823163138,
+      "loss": 0.1652,
+      "step": 15404
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00013315068493150684,
+      "loss": 0.2158,
+      "step": 15405
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0001331382316313823,
+      "loss": 0.2367,
+      "step": 15406
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013312577833125779,
+      "loss": 0.297,
+      "step": 15407
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013311332503113324,
+      "loss": 0.3506,
+      "step": 15408
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0001331008717310087,
+      "loss": 0.1962,
+      "step": 15409
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0001330884184308842,
+      "loss": 0.1754,
+      "step": 15410
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013307596513075962,
+      "loss": 0.1741,
+      "step": 15411
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0001330635118306351,
+      "loss": 0.1715,
+      "step": 15412
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013305105853051057,
+      "loss": 0.1781,
+      "step": 15413
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013303860523038603,
+      "loss": 0.2177,
+      "step": 15414
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0001330261519302615,
+      "loss": 0.2565,
+      "step": 15415
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013301369863013697,
+      "loss": 0.1636,
+      "step": 15416
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013300124533001243,
+      "loss": 0.299,
+      "step": 15417
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013298879202988792,
+      "loss": 0.1439,
+      "step": 15418
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013297633872976338,
+      "loss": 0.1417,
+      "step": 15419
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013296388542963884,
+      "loss": 0.1991,
+      "step": 15420
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013295143212951432,
+      "loss": 0.2713,
+      "step": 15421
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013293897882938978,
+      "loss": 0.1698,
+      "step": 15422
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013292652552926524,
+      "loss": 0.1718,
+      "step": 15423
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013291407222914073,
+      "loss": 0.2213,
+      "step": 15424
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013290161892901619,
+      "loss": 0.2053,
+      "step": 15425
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013288916562889165,
+      "loss": 0.1646,
+      "step": 15426
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0001328767123287671,
+      "loss": 0.2985,
+      "step": 15427
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013286425902864256,
+      "loss": 0.1742,
+      "step": 15428
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013285180572851805,
+      "loss": 0.2728,
+      "step": 15429
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0001328393524283935,
+      "loss": 0.2488,
+      "step": 15430
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013282689912826897,
+      "loss": 0.3108,
+      "step": 15431
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00013281444582814445,
+      "loss": 0.1309,
+      "step": 15432
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013280199252801991,
+      "loss": 0.3864,
+      "step": 15433
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013278953922789537,
+      "loss": 0.218,
+      "step": 15434
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013277708592777086,
+      "loss": 0.2238,
+      "step": 15435
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013276463262764632,
+      "loss": 0.1915,
+      "step": 15436
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013275217932752178,
+      "loss": 0.1862,
+      "step": 15437
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013273972602739726,
+      "loss": 0.1674,
+      "step": 15438
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013272727272727272,
+      "loss": 0.2858,
+      "step": 15439
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013271481942714818,
+      "loss": 0.1752,
+      "step": 15440
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013270236612702367,
+      "loss": 0.1477,
+      "step": 15441
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0001326899128268991,
+      "loss": 0.273,
+      "step": 15442
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0001326774595267746,
+      "loss": 0.1334,
+      "step": 15443
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013266500622665005,
+      "loss": 0.1993,
+      "step": 15444
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0001326525529265255,
+      "loss": 0.3723,
+      "step": 15445
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.000132640099626401,
+      "loss": 0.1994,
+      "step": 15446
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013262764632627645,
+      "loss": 0.2503,
+      "step": 15447
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0001326151930261519,
+      "loss": 0.1553,
+      "step": 15448
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0001326027397260274,
+      "loss": 0.135,
+      "step": 15449
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013259028642590286,
+      "loss": 0.1802,
+      "step": 15450
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013257783312577831,
+      "loss": 0.2617,
+      "step": 15451
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0001325653798256538,
+      "loss": 0.2957,
+      "step": 15452
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013255292652552926,
+      "loss": 0.2341,
+      "step": 15453
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013254047322540472,
+      "loss": 0.1294,
+      "step": 15454
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0001325280199252802,
+      "loss": 0.2017,
+      "step": 15455
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013251556662515566,
+      "loss": 0.1699,
+      "step": 15456
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013250311332503112,
+      "loss": 0.1863,
+      "step": 15457
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00013249066002490658,
+      "loss": 0.307,
+      "step": 15458
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013247820672478204,
+      "loss": 0.2133,
+      "step": 15459
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013246575342465753,
+      "loss": 0.1767,
+      "step": 15460
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.000132453300124533,
+      "loss": 0.1987,
+      "step": 15461
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013244084682440845,
+      "loss": 0.1276,
+      "step": 15462
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013242839352428393,
+      "loss": 0.1466,
+      "step": 15463
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0001324159402241594,
+      "loss": 0.344,
+      "step": 15464
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013240348692403485,
+      "loss": 0.152,
+      "step": 15465
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013239103362391034,
+      "loss": 0.1734,
+      "step": 15466
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0001323785803237858,
+      "loss": 0.1386,
+      "step": 15467
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013236612702366126,
+      "loss": 0.161,
+      "step": 15468
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013235367372353674,
+      "loss": 0.1349,
+      "step": 15469
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0001323412204234122,
+      "loss": 0.3799,
+      "step": 15470
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013232876712328766,
+      "loss": 0.2621,
+      "step": 15471
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013231631382316315,
+      "loss": 0.1685,
+      "step": 15472
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0001323038605230386,
+      "loss": 0.1833,
+      "step": 15473
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013229140722291406,
+      "loss": 0.1105,
+      "step": 15474
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013227895392278952,
+      "loss": 0.2048,
+      "step": 15475
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013226650062266498,
+      "loss": 0.1515,
+      "step": 15476
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013225404732254047,
+      "loss": 0.2802,
+      "step": 15477
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013224159402241593,
+      "loss": 0.1764,
+      "step": 15478
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0001322291407222914,
+      "loss": 0.2416,
+      "step": 15479
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013221668742216687,
+      "loss": 0.1485,
+      "step": 15480
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013220423412204233,
+      "loss": 0.1389,
+      "step": 15481
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0001321917808219178,
+      "loss": 0.2088,
+      "step": 15482
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013217932752179328,
+      "loss": 0.3868,
+      "step": 15483
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013216687422166874,
+      "loss": 0.2312,
+      "step": 15484
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0001321544209215442,
+      "loss": 0.2262,
+      "step": 15485
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013214196762141966,
+      "loss": 0.2275,
+      "step": 15486
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013212951432129514,
+      "loss": 0.2143,
+      "step": 15487
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0001321170610211706,
+      "loss": 0.2206,
+      "step": 15488
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013210460772104606,
+      "loss": 0.2506,
+      "step": 15489
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013209215442092152,
+      "loss": 0.213,
+      "step": 15490
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.000132079701120797,
+      "loss": 0.1347,
+      "step": 15491
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013206724782067247,
+      "loss": 0.1443,
+      "step": 15492
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013205479452054792,
+      "loss": 0.143,
+      "step": 15493
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0001320423412204234,
+      "loss": 0.135,
+      "step": 15494
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013202988792029887,
+      "loss": 0.2695,
+      "step": 15495
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013201743462017433,
+      "loss": 0.2144,
+      "step": 15496
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0001320049813200498,
+      "loss": 0.1463,
+      "step": 15497
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013199252801992527,
+      "loss": 0.1428,
+      "step": 15498
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013198007471980073,
+      "loss": 0.1062,
+      "step": 15499
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0001319676214196762,
+      "loss": 0.1397,
+      "step": 15500
+    },
+    {
+      "epoch": 5.94,
+      "eval_cer": 0.10327955206118189,
+      "eval_loss": 0.37312284111976624,
+      "eval_runtime": 47.3468,
+      "eval_samples_per_second": 37.025,
+      "eval_steps_per_second": 4.647,
+      "eval_wer": 0.16701362175860976,
+      "step": 15500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013195516811955168,
+      "loss": 0.1428,
+      "step": 15501
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013194271481942714,
+      "loss": 0.5106,
+      "step": 15502
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0001319302615193026,
+      "loss": 0.1734,
+      "step": 15503
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013191780821917808,
+      "loss": 0.2702,
+      "step": 15504
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013190535491905354,
+      "loss": 0.1934,
+      "step": 15505
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.000131892901618929,
+      "loss": 0.1834,
+      "step": 15506
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013188044831880446,
+      "loss": 0.244,
+      "step": 15507
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013186799501867992,
+      "loss": 0.3236,
+      "step": 15508
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0001318555417185554,
+      "loss": 0.2373,
+      "step": 15509
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00013184308841843087,
+      "loss": 0.1462,
+      "step": 15510
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013183063511830633,
+      "loss": 0.1658,
+      "step": 15511
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0001318181818181818,
+      "loss": 0.2727,
+      "step": 15512
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013180572851805727,
+      "loss": 0.2092,
+      "step": 15513
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013179327521793273,
+      "loss": 0.3882,
+      "step": 15514
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013178082191780822,
+      "loss": 0.1894,
+      "step": 15515
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013176836861768368,
+      "loss": 0.1863,
+      "step": 15516
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013175591531755913,
+      "loss": 0.1531,
+      "step": 15517
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013174346201743462,
+      "loss": 0.1757,
+      "step": 15518
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013173100871731008,
+      "loss": 0.1239,
+      "step": 15519
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013171855541718554,
+      "loss": 0.2061,
+      "step": 15520
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.000131706102117061,
+      "loss": 0.2063,
+      "step": 15521
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013169364881693646,
+      "loss": 0.2126,
+      "step": 15522
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013168119551681194,
+      "loss": 0.1705,
+      "step": 15523
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0001316687422166874,
+      "loss": 0.1446,
+      "step": 15524
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013165628891656286,
+      "loss": 0.275,
+      "step": 15525
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013164383561643835,
+      "loss": 0.1542,
+      "step": 15526
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0001316313823163138,
+      "loss": 0.2323,
+      "step": 15527
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013161892901618927,
+      "loss": 0.1724,
+      "step": 15528
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013160647571606475,
+      "loss": 0.2275,
+      "step": 15529
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0001315940224159402,
+      "loss": 0.1845,
+      "step": 15530
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013158156911581567,
+      "loss": 0.1836,
+      "step": 15531
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013156911581569116,
+      "loss": 0.1697,
+      "step": 15532
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013155666251556662,
+      "loss": 0.2894,
+      "step": 15533
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013154420921544208,
+      "loss": 0.2497,
+      "step": 15534
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013153175591531756,
+      "loss": 0.1464,
+      "step": 15535
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00013151930261519302,
+      "loss": 0.1572,
+      "step": 15536
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013150684931506848,
+      "loss": 0.1949,
+      "step": 15537
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013149439601494394,
+      "loss": 0.2532,
+      "step": 15538
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0001314819427148194,
+      "loss": 0.3284,
+      "step": 15539
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013146948941469489,
+      "loss": 0.1527,
+      "step": 15540
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013145703611457034,
+      "loss": 0.1593,
+      "step": 15541
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0001314445828144458,
+      "loss": 0.145,
+      "step": 15542
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0001314321295143213,
+      "loss": 0.2389,
+      "step": 15543
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013141967621419675,
+      "loss": 0.1481,
+      "step": 15544
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0001314072229140722,
+      "loss": 0.3036,
+      "step": 15545
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0001313947696139477,
+      "loss": 0.1704,
+      "step": 15546
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013138231631382315,
+      "loss": 0.1959,
+      "step": 15547
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0001313698630136986,
+      "loss": 0.1533,
+      "step": 15548
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0001313574097135741,
+      "loss": 0.1376,
+      "step": 15549
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013134495641344956,
+      "loss": 0.1144,
+      "step": 15550
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013133250311332502,
+      "loss": 0.1616,
+      "step": 15551
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013132004981320048,
+      "loss": 0.3048,
+      "step": 15552
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013130759651307594,
+      "loss": 0.1858,
+      "step": 15553
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013129514321295142,
+      "loss": 0.2101,
+      "step": 15554
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013128268991282688,
+      "loss": 0.1575,
+      "step": 15555
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013127023661270234,
+      "loss": 0.1604,
+      "step": 15556
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013125778331257783,
+      "loss": 0.1396,
+      "step": 15557
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013124533001245329,
+      "loss": 0.3977,
+      "step": 15558
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013123287671232874,
+      "loss": 0.219,
+      "step": 15559
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013122042341220423,
+      "loss": 0.1534,
+      "step": 15560
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0001312079701120797,
+      "loss": 0.2121,
+      "step": 15561
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00013119551681195515,
+      "loss": 0.0943,
+      "step": 15562
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013118306351183064,
+      "loss": 0.2883,
+      "step": 15563
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0001311706102117061,
+      "loss": 0.312,
+      "step": 15564
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013115815691158155,
+      "loss": 0.2352,
+      "step": 15565
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013114570361145704,
+      "loss": 0.2177,
+      "step": 15566
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0001311332503113325,
+      "loss": 0.2359,
+      "step": 15567
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013112079701120796,
+      "loss": 0.2103,
+      "step": 15568
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013110834371108342,
+      "loss": 0.1625,
+      "step": 15569
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013109589041095888,
+      "loss": 0.2806,
+      "step": 15570
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013108343711083436,
+      "loss": 0.2388,
+      "step": 15571
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013107098381070982,
+      "loss": 0.1865,
+      "step": 15572
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013105853051058528,
+      "loss": 0.1971,
+      "step": 15573
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013104607721046077,
+      "loss": 0.1641,
+      "step": 15574
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013103362391033623,
+      "loss": 0.1191,
+      "step": 15575
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013102117061021169,
+      "loss": 0.1819,
+      "step": 15576
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013100871731008717,
+      "loss": 0.2222,
+      "step": 15577
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013099626400996263,
+      "loss": 0.1707,
+      "step": 15578
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0001309838107098381,
+      "loss": 0.098,
+      "step": 15579
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013097135740971358,
+      "loss": 0.1795,
+      "step": 15580
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013095890410958904,
+      "loss": 0.1941,
+      "step": 15581
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0001309464508094645,
+      "loss": 0.2484,
+      "step": 15582
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013093399750933998,
+      "loss": 0.2484,
+      "step": 15583
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013092154420921541,
+      "loss": 0.201,
+      "step": 15584
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0001309090909090909,
+      "loss": 0.1799,
+      "step": 15585
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013089663760896636,
+      "loss": 0.1332,
+      "step": 15586
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00013088418430884182,
+      "loss": 0.1226,
+      "step": 15587
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0001308717310087173,
+      "loss": 0.2431,
+      "step": 15588
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013085927770859276,
+      "loss": 0.2941,
+      "step": 15589
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013084682440846822,
+      "loss": 0.2227,
+      "step": 15590
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0001308343711083437,
+      "loss": 0.1843,
+      "step": 15591
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013082191780821917,
+      "loss": 0.1982,
+      "step": 15592
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013080946450809463,
+      "loss": 0.2336,
+      "step": 15593
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013079701120797011,
+      "loss": 0.1628,
+      "step": 15594
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013078455790784557,
+      "loss": 0.3376,
+      "step": 15595
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013077210460772103,
+      "loss": 0.2259,
+      "step": 15596
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013075965130759652,
+      "loss": 0.1906,
+      "step": 15597
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013074719800747198,
+      "loss": 0.1434,
+      "step": 15598
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013073474470734744,
+      "loss": 0.1245,
+      "step": 15599
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0001307222914072229,
+      "loss": 0.2396,
+      "step": 15600
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013070983810709836,
+      "loss": 0.1743,
+      "step": 15601
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013069738480697384,
+      "loss": 0.2557,
+      "step": 15602
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0001306849315068493,
+      "loss": 0.1531,
+      "step": 15603
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013067247820672476,
+      "loss": 0.2054,
+      "step": 15604
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013066002490660025,
+      "loss": 0.1542,
+      "step": 15605
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0001306475716064757,
+      "loss": 0.1511,
+      "step": 15606
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013063511830635116,
+      "loss": 0.2214,
+      "step": 15607
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013062266500622665,
+      "loss": 0.2262,
+      "step": 15608
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0001306102117061021,
+      "loss": 0.1514,
+      "step": 15609
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013059775840597757,
+      "loss": 0.1974,
+      "step": 15610
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013058530510585306,
+      "loss": 0.2447,
+      "step": 15611
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013057285180572851,
+      "loss": 0.1455,
+      "step": 15612
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013056039850560397,
+      "loss": 0.204,
+      "step": 15613
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00013054794520547946,
+      "loss": 0.2995,
+      "step": 15614
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0001305354919053549,
+      "loss": 0.3121,
+      "step": 15615
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013052303860523038,
+      "loss": 0.1284,
+      "step": 15616
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013051058530510584,
+      "loss": 0.1511,
+      "step": 15617
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0001304981320049813,
+      "loss": 0.2295,
+      "step": 15618
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013048567870485678,
+      "loss": 0.1941,
+      "step": 15619
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013047322540473224,
+      "loss": 0.3972,
+      "step": 15620
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0001304607721046077,
+      "loss": 0.1826,
+      "step": 15621
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0001304483188044832,
+      "loss": 0.1649,
+      "step": 15622
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013043586550435865,
+      "loss": 0.1609,
+      "step": 15623
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0001304234122042341,
+      "loss": 0.1968,
+      "step": 15624
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0001304109589041096,
+      "loss": 0.2022,
+      "step": 15625
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013039850560398505,
+      "loss": 0.3578,
+      "step": 15626
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0001303860523038605,
+      "loss": 0.3529,
+      "step": 15627
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.000130373599003736,
+      "loss": 0.2387,
+      "step": 15628
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013036114570361146,
+      "loss": 0.1761,
+      "step": 15629
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013034869240348692,
+      "loss": 0.1404,
+      "step": 15630
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013033623910336237,
+      "loss": 0.2642,
+      "step": 15631
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013032378580323783,
+      "loss": 0.1256,
+      "step": 15632
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013031133250311332,
+      "loss": 0.3598,
+      "step": 15633
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013029887920298878,
+      "loss": 0.164,
+      "step": 15634
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013028642590286424,
+      "loss": 0.1651,
+      "step": 15635
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013027397260273972,
+      "loss": 0.2177,
+      "step": 15636
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013026151930261518,
+      "loss": 0.1517,
+      "step": 15637
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013024906600249064,
+      "loss": 0.136,
+      "step": 15638
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00013023661270236613,
+      "loss": 0.1716,
+      "step": 15639
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0001302241594022416,
+      "loss": 0.1938,
+      "step": 15640
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013021170610211705,
+      "loss": 0.1704,
+      "step": 15641
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013019925280199253,
+      "loss": 0.2076,
+      "step": 15642
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.000130186799501868,
+      "loss": 0.1533,
+      "step": 15643
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013017434620174345,
+      "loss": 0.1349,
+      "step": 15644
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013016189290161894,
+      "loss": 0.3749,
+      "step": 15645
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013014943960149437,
+      "loss": 0.1891,
+      "step": 15646
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013013698630136986,
+      "loss": 0.1568,
+      "step": 15647
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013012453300124532,
+      "loss": 0.2399,
+      "step": 15648
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013011207970112077,
+      "loss": 0.2723,
+      "step": 15649
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013009962640099626,
+      "loss": 0.1471,
+      "step": 15650
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013008717310087172,
+      "loss": 0.1902,
+      "step": 15651
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013007471980074718,
+      "loss": 0.3234,
+      "step": 15652
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013006226650062267,
+      "loss": 0.1587,
+      "step": 15653
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013004981320049812,
+      "loss": 0.1416,
+      "step": 15654
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013003735990037358,
+      "loss": 0.1506,
+      "step": 15655
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013002490660024907,
+      "loss": 0.188,
+      "step": 15656
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013001245330012453,
+      "loss": 0.123,
+      "step": 15657
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00013,
+      "loss": 0.2325,
+      "step": 15658
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00012998754669987545,
+      "loss": 0.1648,
+      "step": 15659
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00012997509339975093,
+      "loss": 0.2428,
+      "step": 15660
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0001299626400996264,
+      "loss": 0.2579,
+      "step": 15661
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00012995018679950185,
+      "loss": 0.2554,
+      "step": 15662
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0001299377334993773,
+      "loss": 0.1149,
+      "step": 15663
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0001299252801992528,
+      "loss": 0.1202,
+      "step": 15664
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00012991282689912826,
+      "loss": 0.1416,
+      "step": 15665
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00012990037359900372,
+      "loss": 0.2595,
+      "step": 15666
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0001298879202988792,
+      "loss": 0.2312,
+      "step": 15667
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012987546699875466,
+      "loss": 0.1555,
+      "step": 15668
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012986301369863012,
+      "loss": 0.143,
+      "step": 15669
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012985056039850558,
+      "loss": 0.1123,
+      "step": 15670
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012983810709838107,
+      "loss": 0.0859,
+      "step": 15671
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012982565379825653,
+      "loss": 0.1621,
+      "step": 15672
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012981320049813198,
+      "loss": 0.2019,
+      "step": 15673
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012980074719800747,
+      "loss": 0.1625,
+      "step": 15674
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012978829389788293,
+      "loss": 0.1895,
+      "step": 15675
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0001297758405977584,
+      "loss": 0.14,
+      "step": 15676
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012976338729763388,
+      "loss": 0.1923,
+      "step": 15677
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0001297509339975093,
+      "loss": 0.1846,
+      "step": 15678
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0001297384806973848,
+      "loss": 0.1953,
+      "step": 15679
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012972602739726025,
+      "loss": 0.2713,
+      "step": 15680
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0001297135740971357,
+      "loss": 0.2324,
+      "step": 15681
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0001297011207970112,
+      "loss": 0.1357,
+      "step": 15682
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012968866749688666,
+      "loss": 0.147,
+      "step": 15683
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012967621419676212,
+      "loss": 0.1208,
+      "step": 15684
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0001296637608966376,
+      "loss": 0.2075,
+      "step": 15685
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012965130759651306,
+      "loss": 0.3774,
+      "step": 15686
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012963885429638852,
+      "loss": 0.1735,
+      "step": 15687
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.000129626400996264,
+      "loss": 0.2404,
+      "step": 15688
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012961394769613947,
+      "loss": 0.1357,
+      "step": 15689
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012960149439601493,
+      "loss": 0.1245,
+      "step": 15690
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0001295890410958904,
+      "loss": 0.1996,
+      "step": 15691
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012957658779576587,
+      "loss": 0.3598,
+      "step": 15692
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00012956413449564133,
+      "loss": 0.1514,
+      "step": 15693
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001295516811955168,
+      "loss": 0.1948,
+      "step": 15694
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012953922789539225,
+      "loss": 0.1666,
+      "step": 15695
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012952677459526774,
+      "loss": 0.1798,
+      "step": 15696
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001295143212951432,
+      "loss": 0.1246,
+      "step": 15697
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012950186799501865,
+      "loss": 0.2903,
+      "step": 15698
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012948941469489414,
+      "loss": 0.1688,
+      "step": 15699
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001294769613947696,
+      "loss": 0.272,
+      "step": 15700
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012946450809464506,
+      "loss": 0.2276,
+      "step": 15701
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012945205479452054,
+      "loss": 0.0988,
+      "step": 15702
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.000129439601494396,
+      "loss": 0.0944,
+      "step": 15703
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012942714819427146,
+      "loss": 0.308,
+      "step": 15704
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012941469489414695,
+      "loss": 0.1711,
+      "step": 15705
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001294022415940224,
+      "loss": 0.1734,
+      "step": 15706
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012938978829389787,
+      "loss": 0.153,
+      "step": 15707
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012937733499377335,
+      "loss": 0.1389,
+      "step": 15708
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012936488169364879,
+      "loss": 0.1054,
+      "step": 15709
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012935242839352427,
+      "loss": 0.1231,
+      "step": 15710
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012933997509339973,
+      "loss": 0.2892,
+      "step": 15711
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001293275217932752,
+      "loss": 0.2047,
+      "step": 15712
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012931506849315068,
+      "loss": 0.209,
+      "step": 15713
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012930261519302614,
+      "loss": 0.2259,
+      "step": 15714
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0001292901618929016,
+      "loss": 0.0983,
+      "step": 15715
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012927770859277708,
+      "loss": 0.2047,
+      "step": 15716
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012926525529265254,
+      "loss": 0.2171,
+      "step": 15717
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.000129252801992528,
+      "loss": 0.174,
+      "step": 15718
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00012924034869240349,
+      "loss": 0.1951,
+      "step": 15719
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012922789539227894,
+      "loss": 0.1174,
+      "step": 15720
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001292154420921544,
+      "loss": 0.1227,
+      "step": 15721
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001292029887920299,
+      "loss": 0.1228,
+      "step": 15722
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012919053549190535,
+      "loss": 0.2907,
+      "step": 15723
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001291780821917808,
+      "loss": 0.1541,
+      "step": 15724
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012916562889165627,
+      "loss": 0.1411,
+      "step": 15725
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012915317559153173,
+      "loss": 0.1949,
+      "step": 15726
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001291407222914072,
+      "loss": 0.1027,
+      "step": 15727
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012912826899128267,
+      "loss": 0.1799,
+      "step": 15728
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012911581569115813,
+      "loss": 0.3959,
+      "step": 15729
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012910336239103362,
+      "loss": 0.2697,
+      "step": 15730
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012909090909090908,
+      "loss": 0.203,
+      "step": 15731
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012907845579078454,
+      "loss": 0.1571,
+      "step": 15732
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012906600249066002,
+      "loss": 0.1242,
+      "step": 15733
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012905354919053548,
+      "loss": 0.1283,
+      "step": 15734
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012904109589041094,
+      "loss": 0.1375,
+      "step": 15735
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012902864259028643,
+      "loss": 0.2518,
+      "step": 15736
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012901618929016189,
+      "loss": 0.2064,
+      "step": 15737
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012900373599003735,
+      "loss": 0.1519,
+      "step": 15738
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012899128268991283,
+      "loss": 0.1197,
+      "step": 15739
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001289788293897883,
+      "loss": 0.1687,
+      "step": 15740
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012896637608966375,
+      "loss": 0.2414,
+      "step": 15741
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0001289539227895392,
+      "loss": 0.3079,
+      "step": 15742
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012894146948941467,
+      "loss": 0.1437,
+      "step": 15743
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012892901618929015,
+      "loss": 0.2255,
+      "step": 15744
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00012891656288916561,
+      "loss": 0.1343,
+      "step": 15745
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012890410958904107,
+      "loss": 0.143,
+      "step": 15746
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012889165628891656,
+      "loss": 0.1601,
+      "step": 15747
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012887920298879202,
+      "loss": 0.2636,
+      "step": 15748
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012886674968866748,
+      "loss": 0.2273,
+      "step": 15749
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012885429638854296,
+      "loss": 0.1639,
+      "step": 15750
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012884184308841842,
+      "loss": 0.1395,
+      "step": 15751
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012882938978829388,
+      "loss": 0.1341,
+      "step": 15752
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012881693648816937,
+      "loss": 0.1763,
+      "step": 15753
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012880448318804483,
+      "loss": 0.2918,
+      "step": 15754
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001287920298879203,
+      "loss": 0.2928,
+      "step": 15755
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012877957658779575,
+      "loss": 0.1741,
+      "step": 15756
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001287671232876712,
+      "loss": 0.1424,
+      "step": 15757
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001287546699875467,
+      "loss": 0.1507,
+      "step": 15758
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012874221668742215,
+      "loss": 0.1946,
+      "step": 15759
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001287297633872976,
+      "loss": 0.2174,
+      "step": 15760
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001287173100871731,
+      "loss": 0.1969,
+      "step": 15761
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012870485678704856,
+      "loss": 0.1102,
+      "step": 15762
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012869240348692401,
+      "loss": 0.176,
+      "step": 15763
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001286799501867995,
+      "loss": 0.2285,
+      "step": 15764
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012866749688667496,
+      "loss": 0.1224,
+      "step": 15765
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012865504358655042,
+      "loss": 0.1096,
+      "step": 15766
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001286425902864259,
+      "loss": 0.2776,
+      "step": 15767
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012863013698630136,
+      "loss": 0.1386,
+      "step": 15768
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012861768368617682,
+      "loss": 0.1515,
+      "step": 15769
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0001286052303860523,
+      "loss": 0.0914,
+      "step": 15770
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00012859277708592777,
+      "loss": 0.166,
+      "step": 15771
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012858032378580323,
+      "loss": 0.1658,
+      "step": 15772
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001285678704856787,
+      "loss": 0.4071,
+      "step": 15773
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012855541718555415,
+      "loss": 0.1679,
+      "step": 15774
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012854296388542963,
+      "loss": 0.15,
+      "step": 15775
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001285305105853051,
+      "loss": 0.1322,
+      "step": 15776
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012851805728518055,
+      "loss": 0.1846,
+      "step": 15777
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012850560398505604,
+      "loss": 0.2053,
+      "step": 15778
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001284931506849315,
+      "loss": 0.2229,
+      "step": 15779
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012848069738480696,
+      "loss": 0.2129,
+      "step": 15780
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012846824408468244,
+      "loss": 0.2028,
+      "step": 15781
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001284557907845579,
+      "loss": 0.2063,
+      "step": 15782
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012844333748443336,
+      "loss": 0.1701,
+      "step": 15783
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012843088418430885,
+      "loss": 0.1604,
+      "step": 15784
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001284184308841843,
+      "loss": 0.1939,
+      "step": 15785
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012840597758405977,
+      "loss": 0.2186,
+      "step": 15786
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012839352428393525,
+      "loss": 0.2489,
+      "step": 15787
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012838107098381068,
+      "loss": 0.202,
+      "step": 15788
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012836861768368617,
+      "loss": 0.1368,
+      "step": 15789
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012835616438356163,
+      "loss": 0.1588,
+      "step": 15790
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001283437110834371,
+      "loss": 0.2045,
+      "step": 15791
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012833125778331257,
+      "loss": 0.2873,
+      "step": 15792
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012831880448318803,
+      "loss": 0.1534,
+      "step": 15793
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001283063511830635,
+      "loss": 0.1496,
+      "step": 15794
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012829389788293898,
+      "loss": 0.1617,
+      "step": 15795
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00012828144458281444,
+      "loss": 0.1309,
+      "step": 15796
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0001282689912826899,
+      "loss": 0.1317,
+      "step": 15797
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012825653798256538,
+      "loss": 0.34,
+      "step": 15798
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012824408468244084,
+      "loss": 0.1658,
+      "step": 15799
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0001282316313823163,
+      "loss": 0.1478,
+      "step": 15800
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0001282191780821918,
+      "loss": 0.1303,
+      "step": 15801
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012820672478206725,
+      "loss": 0.1335,
+      "step": 15802
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0001281942714819427,
+      "loss": 0.147,
+      "step": 15803
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012818181818181817,
+      "loss": 0.2515,
+      "step": 15804
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012816936488169362,
+      "loss": 0.1841,
+      "step": 15805
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0001281569115815691,
+      "loss": 0.2173,
+      "step": 15806
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012814445828144457,
+      "loss": 0.2155,
+      "step": 15807
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012813200498132003,
+      "loss": 0.1641,
+      "step": 15808
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012811955168119552,
+      "loss": 0.1683,
+      "step": 15809
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012810709838107097,
+      "loss": 0.1805,
+      "step": 15810
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012809464508094643,
+      "loss": 0.208,
+      "step": 15811
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012808219178082192,
+      "loss": 0.1871,
+      "step": 15812
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012806973848069738,
+      "loss": 0.1519,
+      "step": 15813
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012805728518057284,
+      "loss": 0.2405,
+      "step": 15814
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012804483188044832,
+      "loss": 0.1947,
+      "step": 15815
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012803237858032378,
+      "loss": 0.2154,
+      "step": 15816
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012801992528019924,
+      "loss": 0.2335,
+      "step": 15817
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012800747198007473,
+      "loss": 0.2288,
+      "step": 15818
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012799501867995016,
+      "loss": 0.1356,
+      "step": 15819
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012798256537982565,
+      "loss": 0.1955,
+      "step": 15820
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0001279701120797011,
+      "loss": 0.1253,
+      "step": 15821
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012795765877957657,
+      "loss": 0.2953,
+      "step": 15822
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00012794520547945205,
+      "loss": 0.1839,
+      "step": 15823
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001279327521793275,
+      "loss": 0.1671,
+      "step": 15824
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012792029887920297,
+      "loss": 0.1574,
+      "step": 15825
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012790784557907846,
+      "loss": 0.135,
+      "step": 15826
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012789539227895392,
+      "loss": 0.1504,
+      "step": 15827
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012788293897882938,
+      "loss": 0.1432,
+      "step": 15828
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012787048567870486,
+      "loss": 0.2915,
+      "step": 15829
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012785803237858032,
+      "loss": 0.3662,
+      "step": 15830
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012784557907845578,
+      "loss": 0.1235,
+      "step": 15831
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012783312577833124,
+      "loss": 0.1708,
+      "step": 15832
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012782067247820673,
+      "loss": 0.1273,
+      "step": 15833
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012780821917808218,
+      "loss": 0.145,
+      "step": 15834
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012779576587795764,
+      "loss": 0.2484,
+      "step": 15835
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001277833125778331,
+      "loss": 0.2665,
+      "step": 15836
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001277708592777086,
+      "loss": 0.2009,
+      "step": 15837
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012775840597758405,
+      "loss": 0.1669,
+      "step": 15838
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001277459526774595,
+      "loss": 0.2459,
+      "step": 15839
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012773349937733497,
+      "loss": 0.135,
+      "step": 15840
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012772104607721045,
+      "loss": 0.1527,
+      "step": 15841
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0001277085927770859,
+      "loss": 0.154,
+      "step": 15842
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012769613947696137,
+      "loss": 0.211,
+      "step": 15843
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012768368617683686,
+      "loss": 0.2322,
+      "step": 15844
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012767123287671232,
+      "loss": 0.1464,
+      "step": 15845
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012765877957658778,
+      "loss": 0.1633,
+      "step": 15846
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012764632627646326,
+      "loss": 0.1334,
+      "step": 15847
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012763387297633872,
+      "loss": 0.2545,
+      "step": 15848
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.00012762141967621418,
+      "loss": 0.1875,
+      "step": 15849
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012760896637608967,
+      "loss": 0.1795,
+      "step": 15850
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001275965130759651,
+      "loss": 0.18,
+      "step": 15851
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012758405977584059,
+      "loss": 0.1392,
+      "step": 15852
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012757160647571604,
+      "loss": 0.1549,
+      "step": 15853
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001275591531755915,
+      "loss": 0.16,
+      "step": 15854
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.000127546699875467,
+      "loss": 0.1711,
+      "step": 15855
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012753424657534245,
+      "loss": 0.1891,
+      "step": 15856
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001275217932752179,
+      "loss": 0.1678,
+      "step": 15857
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001275093399750934,
+      "loss": 0.1247,
+      "step": 15858
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012749688667496885,
+      "loss": 0.2085,
+      "step": 15859
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001274844333748443,
+      "loss": 0.1359,
+      "step": 15860
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001274719800747198,
+      "loss": 0.3534,
+      "step": 15861
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012745952677459526,
+      "loss": 0.217,
+      "step": 15862
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012744707347447072,
+      "loss": 0.1336,
+      "step": 15863
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001274346201743462,
+      "loss": 0.113,
+      "step": 15864
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012742216687422166,
+      "loss": 0.2307,
+      "step": 15865
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012740971357409712,
+      "loss": 0.2508,
+      "step": 15866
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012739726027397258,
+      "loss": 0.3474,
+      "step": 15867
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012738480697384804,
+      "loss": 0.237,
+      "step": 15868
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012737235367372353,
+      "loss": 0.1557,
+      "step": 15869
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012735990037359899,
+      "loss": 0.148,
+      "step": 15870
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012734744707347444,
+      "loss": 0.1771,
+      "step": 15871
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012733499377334993,
+      "loss": 0.1754,
+      "step": 15872
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0001273225404732254,
+      "loss": 0.285,
+      "step": 15873
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012731008717310085,
+      "loss": 0.2209,
+      "step": 15874
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00012729763387297634,
+      "loss": 0.1454,
+      "step": 15875
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001272851805728518,
+      "loss": 0.1797,
+      "step": 15876
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012727272727272725,
+      "loss": 0.1167,
+      "step": 15877
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012726027397260274,
+      "loss": 0.2061,
+      "step": 15878
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001272478206724782,
+      "loss": 0.221,
+      "step": 15879
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012723536737235366,
+      "loss": 0.2359,
+      "step": 15880
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012722291407222915,
+      "loss": 0.1299,
+      "step": 15881
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012721046077210458,
+      "loss": 0.1908,
+      "step": 15882
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012719800747198006,
+      "loss": 0.1166,
+      "step": 15883
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012718555417185552,
+      "loss": 0.1747,
+      "step": 15884
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012717310087173098,
+      "loss": 0.124,
+      "step": 15885
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012716064757160647,
+      "loss": 0.1979,
+      "step": 15886
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012714819427148193,
+      "loss": 0.1822,
+      "step": 15887
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012713574097135739,
+      "loss": 0.2084,
+      "step": 15888
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012712328767123287,
+      "loss": 0.2285,
+      "step": 15889
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012711083437110833,
+      "loss": 0.2759,
+      "step": 15890
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001270983810709838,
+      "loss": 0.2199,
+      "step": 15891
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012708592777085928,
+      "loss": 0.2498,
+      "step": 15892
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012707347447073474,
+      "loss": 0.1992,
+      "step": 15893
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001270610211706102,
+      "loss": 0.1955,
+      "step": 15894
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012704856787048568,
+      "loss": 0.2185,
+      "step": 15895
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012703611457036114,
+      "loss": 0.1432,
+      "step": 15896
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001270236612702366,
+      "loss": 0.1989,
+      "step": 15897
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012701120797011206,
+      "loss": 0.2738,
+      "step": 15898
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012699875466998752,
+      "loss": 0.2204,
+      "step": 15899
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.000126986301369863,
+      "loss": 0.2179,
+      "step": 15900
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00012697384806973846,
+      "loss": 0.1497,
+      "step": 15901
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012696139476961392,
+      "loss": 0.1026,
+      "step": 15902
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0001269489414694894,
+      "loss": 0.1303,
+      "step": 15903
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012693648816936487,
+      "loss": 0.1502,
+      "step": 15904
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012692403486924033,
+      "loss": 0.2622,
+      "step": 15905
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012691158156911581,
+      "loss": 0.194,
+      "step": 15906
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012689912826899127,
+      "loss": 0.1114,
+      "step": 15907
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012688667496886673,
+      "loss": 0.1565,
+      "step": 15908
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012687422166874222,
+      "loss": 0.1323,
+      "step": 15909
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012686176836861768,
+      "loss": 0.1495,
+      "step": 15910
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012684931506849314,
+      "loss": 0.2151,
+      "step": 15911
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012683686176836862,
+      "loss": 0.1362,
+      "step": 15912
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012682440846824406,
+      "loss": 0.1307,
+      "step": 15913
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012681195516811954,
+      "loss": 0.2132,
+      "step": 15914
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.000126799501867995,
+      "loss": 0.1804,
+      "step": 15915
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012678704856787046,
+      "loss": 0.1326,
+      "step": 15916
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012677459526774595,
+      "loss": 0.1942,
+      "step": 15917
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0001267621419676214,
+      "loss": 0.1987,
+      "step": 15918
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012674968866749686,
+      "loss": 0.2015,
+      "step": 15919
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012673723536737235,
+      "loss": 0.1298,
+      "step": 15920
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0001267247820672478,
+      "loss": 0.118,
+      "step": 15921
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012671232876712327,
+      "loss": 0.1227,
+      "step": 15922
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012669987546699876,
+      "loss": 0.2276,
+      "step": 15923
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012668742216687421,
+      "loss": 0.1646,
+      "step": 15924
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012667496886674967,
+      "loss": 0.2317,
+      "step": 15925
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012666251556662516,
+      "loss": 0.1123,
+      "step": 15926
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00012665006226650062,
+      "loss": 0.094,
+      "step": 15927
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012663760896637608,
+      "loss": 0.1357,
+      "step": 15928
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012662515566625154,
+      "loss": 0.2544,
+      "step": 15929
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.000126612702366127,
+      "loss": 0.2419,
+      "step": 15930
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012660024906600248,
+      "loss": 0.1638,
+      "step": 15931
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012658779576587794,
+      "loss": 0.1729,
+      "step": 15932
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001265753424657534,
+      "loss": 0.1594,
+      "step": 15933
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001265628891656289,
+      "loss": 0.1203,
+      "step": 15934
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012655043586550435,
+      "loss": 0.1837,
+      "step": 15935
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001265379825653798,
+      "loss": 0.2639,
+      "step": 15936
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001265255292652553,
+      "loss": 0.2355,
+      "step": 15937
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012651307596513075,
+      "loss": 0.1693,
+      "step": 15938
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001265006226650062,
+      "loss": 0.1699,
+      "step": 15939
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001264881693648817,
+      "loss": 0.1705,
+      "step": 15940
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012647571606475716,
+      "loss": 0.1153,
+      "step": 15941
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012646326276463262,
+      "loss": 0.185,
+      "step": 15942
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001264508094645081,
+      "loss": 0.2264,
+      "step": 15943
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012643835616438356,
+      "loss": 0.2037,
+      "step": 15944
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012642590286425902,
+      "loss": 0.148,
+      "step": 15945
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012641344956413448,
+      "loss": 0.1953,
+      "step": 15946
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012640099626400994,
+      "loss": 0.1013,
+      "step": 15947
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012638854296388542,
+      "loss": 0.1666,
+      "step": 15948
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012637608966376088,
+      "loss": 0.1425,
+      "step": 15949
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012636363636363634,
+      "loss": 0.1746,
+      "step": 15950
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012635118306351183,
+      "loss": 0.1916,
+      "step": 15951
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0001263387297633873,
+      "loss": 0.1902,
+      "step": 15952
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012632627646326275,
+      "loss": 0.1117,
+      "step": 15953
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.00012631382316313823,
+      "loss": 0.3601,
+      "step": 15954
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001263013698630137,
+      "loss": 0.2194,
+      "step": 15955
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012628891656288915,
+      "loss": 0.2109,
+      "step": 15956
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012627646326276464,
+      "loss": 0.2312,
+      "step": 15957
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001262640099626401,
+      "loss": 0.1464,
+      "step": 15958
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012625155666251556,
+      "loss": 0.2166,
+      "step": 15959
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012623910336239104,
+      "loss": 0.0966,
+      "step": 15960
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012622665006226647,
+      "loss": 0.2571,
+      "step": 15961
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012621419676214196,
+      "loss": 0.1768,
+      "step": 15962
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012620174346201742,
+      "loss": 0.2109,
+      "step": 15963
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012618929016189288,
+      "loss": 0.1445,
+      "step": 15964
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012617683686176837,
+      "loss": 0.1573,
+      "step": 15965
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012616438356164382,
+      "loss": 0.1615,
+      "step": 15966
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012615193026151928,
+      "loss": 0.2442,
+      "step": 15967
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012613947696139477,
+      "loss": 0.178,
+      "step": 15968
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012612702366127023,
+      "loss": 0.2056,
+      "step": 15969
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001261145703611457,
+      "loss": 0.2155,
+      "step": 15970
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012610211706102117,
+      "loss": 0.0993,
+      "step": 15971
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012608966376089663,
+      "loss": 0.2007,
+      "step": 15972
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001260772104607721,
+      "loss": 0.1971,
+      "step": 15973
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012606475716064758,
+      "loss": 0.2667,
+      "step": 15974
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012605230386052304,
+      "loss": 0.1391,
+      "step": 15975
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001260398505603985,
+      "loss": 0.153,
+      "step": 15976
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012602739726027396,
+      "loss": 0.1958,
+      "step": 15977
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012601494396014942,
+      "loss": 0.2018,
+      "step": 15978
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0001260024906600249,
+      "loss": 0.2118,
+      "step": 15979
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00012599003735990036,
+      "loss": 0.265,
+      "step": 15980
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012597758405977582,
+      "loss": 0.1415,
+      "step": 15981
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001259651307596513,
+      "loss": 0.2302,
+      "step": 15982
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012595267745952677,
+      "loss": 0.1263,
+      "step": 15983
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012594022415940223,
+      "loss": 0.1064,
+      "step": 15984
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001259277708592777,
+      "loss": 0.2496,
+      "step": 15985
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012591531755915317,
+      "loss": 0.2581,
+      "step": 15986
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012590286425902863,
+      "loss": 0.2033,
+      "step": 15987
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012589041095890412,
+      "loss": 0.1474,
+      "step": 15988
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012587795765877958,
+      "loss": 0.1573,
+      "step": 15989
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012586550435865503,
+      "loss": 0.1425,
+      "step": 15990
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001258530510585305,
+      "loss": 0.1066,
+      "step": 15991
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012584059775840595,
+      "loss": 0.2682,
+      "step": 15992
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012582814445828144,
+      "loss": 0.2103,
+      "step": 15993
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001258156911581569,
+      "loss": 0.1848,
+      "step": 15994
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012580323785803236,
+      "loss": 0.22,
+      "step": 15995
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012579078455790784,
+      "loss": 0.1804,
+      "step": 15996
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001257783312577833,
+      "loss": 0.1339,
+      "step": 15997
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012576587795765876,
+      "loss": 0.2224,
+      "step": 15998
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012575342465753425,
+      "loss": 0.1945,
+      "step": 15999
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001257409713574097,
+      "loss": 0.2056,
+      "step": 16000
+    },
+    {
+      "epoch": 6.13,
+      "eval_cer": 0.10304153579023353,
+      "eval_loss": 0.37824106216430664,
+      "eval_runtime": 47.756,
+      "eval_samples_per_second": 36.707,
+      "eval_steps_per_second": 4.607,
+      "eval_wer": 0.1650450287369326,
+      "step": 16000
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012572851805728517,
+      "loss": 0.1576,
+      "step": 16001
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012571606475716063,
+      "loss": 0.1885,
+      "step": 16002
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0001257036114570361,
+      "loss": 0.1006,
+      "step": 16003
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012569115815691157,
+      "loss": 0.2463,
+      "step": 16004
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012567870485678703,
+      "loss": 0.2265,
+      "step": 16005
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00012566625155666252,
+      "loss": 0.1976,
+      "step": 16006
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012565379825653798,
+      "loss": 0.1371,
+      "step": 16007
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012564134495641344,
+      "loss": 0.2233,
+      "step": 16008
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001256288916562889,
+      "loss": 0.2018,
+      "step": 16009
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012561643835616438,
+      "loss": 0.2177,
+      "step": 16010
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012560398505603984,
+      "loss": 0.1854,
+      "step": 16011
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001255915317559153,
+      "loss": 0.1397,
+      "step": 16012
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012557907845579076,
+      "loss": 0.1524,
+      "step": 16013
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012556662515566624,
+      "loss": 0.2552,
+      "step": 16014
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001255541718555417,
+      "loss": 0.114,
+      "step": 16015
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012554171855541716,
+      "loss": 0.1189,
+      "step": 16016
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012552926525529265,
+      "loss": 0.2926,
+      "step": 16017
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001255168119551681,
+      "loss": 0.1818,
+      "step": 16018
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012550435865504357,
+      "loss": 0.1609,
+      "step": 16019
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012549190535491905,
+      "loss": 0.2169,
+      "step": 16020
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001254794520547945,
+      "loss": 0.1288,
+      "step": 16021
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012546699875466997,
+      "loss": 0.1748,
+      "step": 16022
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012545454545454543,
+      "loss": 0.276,
+      "step": 16023
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001254420921544209,
+      "loss": 0.3374,
+      "step": 16024
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012542963885429638,
+      "loss": 0.1466,
+      "step": 16025
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012541718555417184,
+      "loss": 0.1697,
+      "step": 16026
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001254047322540473,
+      "loss": 0.1436,
+      "step": 16027
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012539227895392278,
+      "loss": 0.1373,
+      "step": 16028
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012537982565379824,
+      "loss": 0.3157,
+      "step": 16029
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0001253673723536737,
+      "loss": 0.1615,
+      "step": 16030
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012535491905354919,
+      "loss": 0.1456,
+      "step": 16031
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00012534246575342465,
+      "loss": 0.1257,
+      "step": 16032
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001253300124533001,
+      "loss": 0.1774,
+      "step": 16033
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001253175591531756,
+      "loss": 0.1308,
+      "step": 16034
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012530510585305105,
+      "loss": 0.1424,
+      "step": 16035
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001252926525529265,
+      "loss": 0.2855,
+      "step": 16036
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.000125280199252802,
+      "loss": 0.2705,
+      "step": 16037
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012526774595267745,
+      "loss": 0.3175,
+      "step": 16038
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001252552926525529,
+      "loss": 0.107,
+      "step": 16039
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012524283935242837,
+      "loss": 0.134,
+      "step": 16040
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012523038605230383,
+      "loss": 0.1745,
+      "step": 16041
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012521793275217932,
+      "loss": 0.233,
+      "step": 16042
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012520547945205478,
+      "loss": 0.2419,
+      "step": 16043
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012519302615193024,
+      "loss": 0.1438,
+      "step": 16044
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012518057285180572,
+      "loss": 0.123,
+      "step": 16045
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012516811955168118,
+      "loss": 0.2237,
+      "step": 16046
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012515566625155664,
+      "loss": 0.1467,
+      "step": 16047
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012514321295143213,
+      "loss": 0.2123,
+      "step": 16048
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0001251307596513076,
+      "loss": 0.1773,
+      "step": 16049
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012511830635118305,
+      "loss": 0.1674,
+      "step": 16050
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012510585305105853,
+      "loss": 0.2322,
+      "step": 16051
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.000125093399750934,
+      "loss": 0.1588,
+      "step": 16052
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012508094645080945,
+      "loss": 0.1316,
+      "step": 16053
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012506849315068494,
+      "loss": 0.3496,
+      "step": 16054
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012505603985056037,
+      "loss": 0.1689,
+      "step": 16055
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012504358655043585,
+      "loss": 0.1708,
+      "step": 16056
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012503113325031131,
+      "loss": 0.1844,
+      "step": 16057
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00012501867995018677,
+      "loss": 0.1996,
+      "step": 16058
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012500622665006226,
+      "loss": 0.1297,
+      "step": 16059
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012499377334993772,
+      "loss": 0.1017,
+      "step": 16060
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012498132004981318,
+      "loss": 0.2616,
+      "step": 16061
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012496886674968866,
+      "loss": 0.1269,
+      "step": 16062
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012495641344956412,
+      "loss": 0.1558,
+      "step": 16063
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012494396014943958,
+      "loss": 0.1221,
+      "step": 16064
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012493150684931507,
+      "loss": 0.1034,
+      "step": 16065
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012491905354919053,
+      "loss": 0.1721,
+      "step": 16066
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.000124906600249066,
+      "loss": 0.2023,
+      "step": 16067
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012489414694894147,
+      "loss": 0.1704,
+      "step": 16068
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012488169364881693,
+      "loss": 0.1675,
+      "step": 16069
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001248692403486924,
+      "loss": 0.1259,
+      "step": 16070
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012485678704856785,
+      "loss": 0.1747,
+      "step": 16071
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001248443337484433,
+      "loss": 0.1201,
+      "step": 16072
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001248318804483188,
+      "loss": 0.2795,
+      "step": 16073
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012481942714819426,
+      "loss": 0.2254,
+      "step": 16074
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012480697384806971,
+      "loss": 0.1215,
+      "step": 16075
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001247945205479452,
+      "loss": 0.1789,
+      "step": 16076
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012478206724782066,
+      "loss": 0.1482,
+      "step": 16077
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012476961394769612,
+      "loss": 0.0957,
+      "step": 16078
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0001247571606475716,
+      "loss": 0.1871,
+      "step": 16079
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012474470734744706,
+      "loss": 0.2193,
+      "step": 16080
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012473225404732252,
+      "loss": 0.1513,
+      "step": 16081
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.000124719800747198,
+      "loss": 0.1466,
+      "step": 16082
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012470734744707347,
+      "loss": 0.1728,
+      "step": 16083
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00012469489414694893,
+      "loss": 0.1765,
+      "step": 16084
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012468244084682441,
+      "loss": 0.2966,
+      "step": 16085
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012466998754669985,
+      "loss": 0.2193,
+      "step": 16086
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012465753424657533,
+      "loss": 0.1827,
+      "step": 16087
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001246450809464508,
+      "loss": 0.1458,
+      "step": 16088
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012463262764632625,
+      "loss": 0.1762,
+      "step": 16089
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012462017434620174,
+      "loss": 0.1199,
+      "step": 16090
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001246077210460772,
+      "loss": 0.1695,
+      "step": 16091
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012459526774595266,
+      "loss": 0.3074,
+      "step": 16092
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012458281444582814,
+      "loss": 0.169,
+      "step": 16093
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001245703611457036,
+      "loss": 0.1219,
+      "step": 16094
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012455790784557906,
+      "loss": 0.1419,
+      "step": 16095
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012454545454545455,
+      "loss": 0.2373,
+      "step": 16096
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012453300124533,
+      "loss": 0.1448,
+      "step": 16097
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012452054794520547,
+      "loss": 0.3033,
+      "step": 16098
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012450809464508095,
+      "loss": 0.2086,
+      "step": 16099
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001244956413449564,
+      "loss": 0.198,
+      "step": 16100
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012448318804483187,
+      "loss": 0.119,
+      "step": 16101
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012447073474470733,
+      "loss": 0.147,
+      "step": 16102
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001244582814445828,
+      "loss": 0.1593,
+      "step": 16103
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012444582814445827,
+      "loss": 0.2945,
+      "step": 16104
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012443337484433373,
+      "loss": 0.1994,
+      "step": 16105
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001244209215442092,
+      "loss": 0.164,
+      "step": 16106
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012440846824408468,
+      "loss": 0.1514,
+      "step": 16107
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012439601494396014,
+      "loss": 0.1581,
+      "step": 16108
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0001243835616438356,
+      "loss": 0.1133,
+      "step": 16109
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00012437110834371108,
+      "loss": 0.1304,
+      "step": 16110
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012435865504358654,
+      "loss": 0.1619,
+      "step": 16111
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.000124346201743462,
+      "loss": 0.1822,
+      "step": 16112
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001243337484433375,
+      "loss": 0.2563,
+      "step": 16113
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012432129514321295,
+      "loss": 0.1769,
+      "step": 16114
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001243088418430884,
+      "loss": 0.1446,
+      "step": 16115
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001242963885429639,
+      "loss": 0.1779,
+      "step": 16116
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012428393524283935,
+      "loss": 0.166,
+      "step": 16117
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001242714819427148,
+      "loss": 0.2258,
+      "step": 16118
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012425902864259027,
+      "loss": 0.1899,
+      "step": 16119
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012424657534246573,
+      "loss": 0.1982,
+      "step": 16120
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012423412204234122,
+      "loss": 0.2509,
+      "step": 16121
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012422166874221667,
+      "loss": 0.261,
+      "step": 16122
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012420921544209213,
+      "loss": 0.2018,
+      "step": 16123
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012419676214196762,
+      "loss": 0.2438,
+      "step": 16124
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012418430884184308,
+      "loss": 0.1798,
+      "step": 16125
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012417185554171854,
+      "loss": 0.176,
+      "step": 16126
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012415940224159403,
+      "loss": 0.143,
+      "step": 16127
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012414694894146948,
+      "loss": 0.1827,
+      "step": 16128
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012413449564134494,
+      "loss": 0.2378,
+      "step": 16129
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012412204234122043,
+      "loss": 0.2147,
+      "step": 16130
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001241095890410959,
+      "loss": 0.1734,
+      "step": 16131
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012409713574097135,
+      "loss": 0.1415,
+      "step": 16132
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001240846824408468,
+      "loss": 0.2032,
+      "step": 16133
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012407222914072227,
+      "loss": 0.1297,
+      "step": 16134
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00012405977584059775,
+      "loss": 0.251,
+      "step": 16135
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0001240473225404732,
+      "loss": 0.2863,
+      "step": 16136
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012403486924034867,
+      "loss": 0.1513,
+      "step": 16137
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012402241594022416,
+      "loss": 0.1217,
+      "step": 16138
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012400996264009962,
+      "loss": 0.2096,
+      "step": 16139
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012399750933997508,
+      "loss": 0.117,
+      "step": 16140
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012398505603985056,
+      "loss": 0.199,
+      "step": 16141
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012397260273972602,
+      "loss": 0.2318,
+      "step": 16142
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012396014943960148,
+      "loss": 0.1699,
+      "step": 16143
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012394769613947697,
+      "loss": 0.1627,
+      "step": 16144
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012393524283935243,
+      "loss": 0.1774,
+      "step": 16145
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012392278953922788,
+      "loss": 0.1158,
+      "step": 16146
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012391033623910337,
+      "loss": 0.2052,
+      "step": 16147
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012389788293897883,
+      "loss": 0.2423,
+      "step": 16148
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001238854296388543,
+      "loss": 0.1624,
+      "step": 16149
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012387297633872975,
+      "loss": 0.1594,
+      "step": 16150
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001238605230386052,
+      "loss": 0.2053,
+      "step": 16151
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001238480697384807,
+      "loss": 0.2089,
+      "step": 16152
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012383561643835615,
+      "loss": 0.2006,
+      "step": 16153
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001238231631382316,
+      "loss": 0.2533,
+      "step": 16154
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001238107098381071,
+      "loss": 0.2336,
+      "step": 16155
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012379825653798256,
+      "loss": 0.2048,
+      "step": 16156
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012378580323785802,
+      "loss": 0.1142,
+      "step": 16157
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001237733499377335,
+      "loss": 0.2649,
+      "step": 16158
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012376089663760896,
+      "loss": 0.1442,
+      "step": 16159
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012374844333748442,
+      "loss": 0.1773,
+      "step": 16160
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0001237359900373599,
+      "loss": 0.3356,
+      "step": 16161
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012372353673723537,
+      "loss": 0.1537,
+      "step": 16162
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012371108343711083,
+      "loss": 0.2154,
+      "step": 16163
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012369863013698629,
+      "loss": 0.2332,
+      "step": 16164
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012368617683686174,
+      "loss": 0.1083,
+      "step": 16165
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012367372353673723,
+      "loss": 0.2197,
+      "step": 16166
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001236612702366127,
+      "loss": 0.1997,
+      "step": 16167
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012364881693648815,
+      "loss": 0.2233,
+      "step": 16168
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012363636363636364,
+      "loss": 0.1493,
+      "step": 16169
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001236239103362391,
+      "loss": 0.1418,
+      "step": 16170
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012361145703611455,
+      "loss": 0.1153,
+      "step": 16171
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012359900373599004,
+      "loss": 0.1956,
+      "step": 16172
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001235865504358655,
+      "loss": 0.2937,
+      "step": 16173
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012357409713574096,
+      "loss": 0.1612,
+      "step": 16174
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012356164383561642,
+      "loss": 0.1355,
+      "step": 16175
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001235491905354919,
+      "loss": 0.1598,
+      "step": 16176
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012353673723536736,
+      "loss": 0.1282,
+      "step": 16177
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012352428393524282,
+      "loss": 0.1623,
+      "step": 16178
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001235118306351183,
+      "loss": 0.3034,
+      "step": 16179
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012349937733499377,
+      "loss": 0.2338,
+      "step": 16180
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012348692403486923,
+      "loss": 0.1866,
+      "step": 16181
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012347447073474469,
+      "loss": 0.1499,
+      "step": 16182
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012346201743462015,
+      "loss": 0.1521,
+      "step": 16183
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012344956413449563,
+      "loss": 0.1192,
+      "step": 16184
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001234371108343711,
+      "loss": 0.1788,
+      "step": 16185
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012342465753424655,
+      "loss": 0.3125,
+      "step": 16186
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00012341220423412204,
+      "loss": 0.2411,
+      "step": 16187
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0001233997509339975,
+      "loss": 0.2198,
+      "step": 16188
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012338729763387295,
+      "loss": 0.209,
+      "step": 16189
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012337484433374844,
+      "loss": 0.138,
+      "step": 16190
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001233623910336239,
+      "loss": 0.158,
+      "step": 16191
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012334993773349936,
+      "loss": 0.2339,
+      "step": 16192
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012333748443337485,
+      "loss": 0.1412,
+      "step": 16193
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001233250311332503,
+      "loss": 0.2326,
+      "step": 16194
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012331257783312576,
+      "loss": 0.1608,
+      "step": 16195
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012330012453300122,
+      "loss": 0.1933,
+      "step": 16196
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012328767123287668,
+      "loss": 0.1618,
+      "step": 16197
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012327521793275217,
+      "loss": 0.1968,
+      "step": 16198
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012326276463262763,
+      "loss": 0.1836,
+      "step": 16199
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001232503113325031,
+      "loss": 0.2436,
+      "step": 16200
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012323785803237857,
+      "loss": 0.2006,
+      "step": 16201
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012322540473225403,
+      "loss": 0.1515,
+      "step": 16202
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001232129514321295,
+      "loss": 0.1621,
+      "step": 16203
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012320049813200498,
+      "loss": 0.1583,
+      "step": 16204
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012318804483188044,
+      "loss": 0.1848,
+      "step": 16205
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001231755915317559,
+      "loss": 0.1709,
+      "step": 16206
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012316313823163138,
+      "loss": 0.1165,
+      "step": 16207
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012315068493150684,
+      "loss": 0.1585,
+      "step": 16208
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001231382316313823,
+      "loss": 0.1675,
+      "step": 16209
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001231257783312578,
+      "loss": 0.1981,
+      "step": 16210
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012311332503113325,
+      "loss": 0.2428,
+      "step": 16211
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0001231008717310087,
+      "loss": 0.2027,
+      "step": 16212
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012308841843088416,
+      "loss": 0.1371,
+      "step": 16213
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00012307596513075962,
+      "loss": 0.1601,
+      "step": 16214
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0001230635118306351,
+      "loss": 0.2166,
+      "step": 16215
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012305105853051057,
+      "loss": 0.1198,
+      "step": 16216
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012303860523038603,
+      "loss": 0.3377,
+      "step": 16217
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012302615193026151,
+      "loss": 0.2107,
+      "step": 16218
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012301369863013697,
+      "loss": 0.1616,
+      "step": 16219
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012300124533001243,
+      "loss": 0.1423,
+      "step": 16220
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012298879202988792,
+      "loss": 0.1244,
+      "step": 16221
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012297633872976338,
+      "loss": 0.1709,
+      "step": 16222
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012296388542963884,
+      "loss": 0.2223,
+      "step": 16223
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012295143212951432,
+      "loss": 0.2383,
+      "step": 16224
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012293897882938978,
+      "loss": 0.1358,
+      "step": 16225
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012292652552926524,
+      "loss": 0.1474,
+      "step": 16226
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012291407222914073,
+      "loss": 0.2069,
+      "step": 16227
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012290161892901616,
+      "loss": 0.2089,
+      "step": 16228
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012288916562889165,
+      "loss": 0.2988,
+      "step": 16229
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0001228767123287671,
+      "loss": 0.246,
+      "step": 16230
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012286425902864256,
+      "loss": 0.295,
+      "step": 16231
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012285180572851805,
+      "loss": 0.2567,
+      "step": 16232
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0001228393524283935,
+      "loss": 0.1658,
+      "step": 16233
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012282689912826897,
+      "loss": 0.1252,
+      "step": 16234
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012281444582814446,
+      "loss": 0.1858,
+      "step": 16235
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012280199252801991,
+      "loss": 0.2208,
+      "step": 16236
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012278953922789537,
+      "loss": 0.2223,
+      "step": 16237
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012277708592777086,
+      "loss": 0.1304,
+      "step": 16238
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012276463262764632,
+      "loss": 0.215,
+      "step": 16239
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012275217932752178,
+      "loss": 0.1164,
+      "step": 16240
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00012273972602739726,
+      "loss": 0.1565,
+      "step": 16241
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012272727272727272,
+      "loss": 0.1981,
+      "step": 16242
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012271481942714818,
+      "loss": 0.1826,
+      "step": 16243
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012270236612702364,
+      "loss": 0.1599,
+      "step": 16244
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001226899128268991,
+      "loss": 0.1557,
+      "step": 16245
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001226774595267746,
+      "loss": 0.2051,
+      "step": 16246
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012266500622665005,
+      "loss": 0.2514,
+      "step": 16247
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001226525529265255,
+      "loss": 0.2689,
+      "step": 16248
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.000122640099626401,
+      "loss": 0.2143,
+      "step": 16249
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012262764632627645,
+      "loss": 0.1737,
+      "step": 16250
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001226151930261519,
+      "loss": 0.1419,
+      "step": 16251
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001226027397260274,
+      "loss": 0.1601,
+      "step": 16252
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012259028642590286,
+      "loss": 0.1812,
+      "step": 16253
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012257783312577832,
+      "loss": 0.2042,
+      "step": 16254
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001225653798256538,
+      "loss": 0.1941,
+      "step": 16255
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012255292652552926,
+      "loss": 0.2275,
+      "step": 16256
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012254047322540472,
+      "loss": 0.1157,
+      "step": 16257
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001225280199252802,
+      "loss": 0.1208,
+      "step": 16258
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012251556662515564,
+      "loss": 0.1961,
+      "step": 16259
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012250311332503112,
+      "loss": 0.1857,
+      "step": 16260
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012249066002490658,
+      "loss": 0.2976,
+      "step": 16261
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012247820672478204,
+      "loss": 0.2408,
+      "step": 16262
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012246575342465753,
+      "loss": 0.2547,
+      "step": 16263
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.000122453300124533,
+      "loss": 0.2101,
+      "step": 16264
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012244084682440845,
+      "loss": 0.1498,
+      "step": 16265
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00012242839352428393,
+      "loss": 0.1287,
+      "step": 16266
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0001224159402241594,
+      "loss": 0.4186,
+      "step": 16267
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012240348692403485,
+      "loss": 0.2648,
+      "step": 16268
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012239103362391034,
+      "loss": 0.142,
+      "step": 16269
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001223785803237858,
+      "loss": 0.1833,
+      "step": 16270
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012236612702366126,
+      "loss": 0.2369,
+      "step": 16271
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012235367372353674,
+      "loss": 0.1746,
+      "step": 16272
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001223412204234122,
+      "loss": 0.4636,
+      "step": 16273
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012232876712328766,
+      "loss": 0.1769,
+      "step": 16274
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012231631382316312,
+      "loss": 0.1497,
+      "step": 16275
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012230386052303858,
+      "loss": 0.1534,
+      "step": 16276
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012229140722291407,
+      "loss": 0.1812,
+      "step": 16277
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012227895392278953,
+      "loss": 0.2001,
+      "step": 16278
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012226650062266498,
+      "loss": 0.1764,
+      "step": 16279
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012225404732254047,
+      "loss": 0.1683,
+      "step": 16280
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012224159402241593,
+      "loss": 0.1437,
+      "step": 16281
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001222291407222914,
+      "loss": 0.2075,
+      "step": 16282
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012221668742216688,
+      "loss": 0.1692,
+      "step": 16283
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012220423412204233,
+      "loss": 0.1459,
+      "step": 16284
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001221917808219178,
+      "loss": 0.1926,
+      "step": 16285
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012217932752179328,
+      "loss": 0.2613,
+      "step": 16286
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012216687422166874,
+      "loss": 0.1809,
+      "step": 16287
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001221544209215442,
+      "loss": 0.1381,
+      "step": 16288
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012214196762141968,
+      "loss": 0.1235,
+      "step": 16289
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012212951432129512,
+      "loss": 0.1364,
+      "step": 16290
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0001221170610211706,
+      "loss": 0.1719,
+      "step": 16291
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012210460772104606,
+      "loss": 0.3947,
+      "step": 16292
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00012209215442092152,
+      "loss": 0.1781,
+      "step": 16293
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.000122079701120797,
+      "loss": 0.1544,
+      "step": 16294
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012206724782067247,
+      "loss": 0.1617,
+      "step": 16295
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012205479452054794,
+      "loss": 0.1169,
+      "step": 16296
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001220423412204234,
+      "loss": 0.0971,
+      "step": 16297
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012202988792029887,
+      "loss": 0.3002,
+      "step": 16298
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012201743462017433,
+      "loss": 0.3242,
+      "step": 16299
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001220049813200498,
+      "loss": 0.1585,
+      "step": 16300
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012199252801992528,
+      "loss": 0.1034,
+      "step": 16301
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012198007471980073,
+      "loss": 0.2693,
+      "step": 16302
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012196762141967621,
+      "loss": 0.1441,
+      "step": 16303
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012195516811955168,
+      "loss": 0.2375,
+      "step": 16304
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012194271481942714,
+      "loss": 0.1793,
+      "step": 16305
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0001219302615193026,
+      "loss": 0.1621,
+      "step": 16306
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012191780821917807,
+      "loss": 0.1133,
+      "step": 16307
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012190535491905353,
+      "loss": 0.1362,
+      "step": 16308
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.000121892901618929,
+      "loss": 0.1667,
+      "step": 16309
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012188044831880446,
+      "loss": 0.1364,
+      "step": 16310
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012186799501867994,
+      "loss": 0.2783,
+      "step": 16311
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012185554171855541,
+      "loss": 0.1605,
+      "step": 16312
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012184308841843087,
+      "loss": 0.1287,
+      "step": 16313
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012183063511830634,
+      "loss": 0.1949,
+      "step": 16314
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012181818181818181,
+      "loss": 0.1141,
+      "step": 16315
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012180572851805727,
+      "loss": 0.1326,
+      "step": 16316
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012179327521793274,
+      "loss": 0.2192,
+      "step": 16317
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012178082191780822,
+      "loss": 0.1504,
+      "step": 16318
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.00012176836861768368,
+      "loss": 0.2461,
+      "step": 16319
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012175591531755915,
+      "loss": 0.1234,
+      "step": 16320
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012174346201743462,
+      "loss": 0.1151,
+      "step": 16321
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012173100871731007,
+      "loss": 0.1359,
+      "step": 16322
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012171855541718554,
+      "loss": 0.2785,
+      "step": 16323
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.000121706102117061,
+      "loss": 0.115,
+      "step": 16324
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012169364881693647,
+      "loss": 0.1073,
+      "step": 16325
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012168119551681194,
+      "loss": 0.1058,
+      "step": 16326
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0001216687422166874,
+      "loss": 0.1238,
+      "step": 16327
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012165628891656288,
+      "loss": 0.154,
+      "step": 16328
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012164383561643835,
+      "loss": 0.3054,
+      "step": 16329
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012163138231631381,
+      "loss": 0.1629,
+      "step": 16330
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012161892901618928,
+      "loss": 0.1405,
+      "step": 16331
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012160647571606475,
+      "loss": 0.1208,
+      "step": 16332
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012159402241594021,
+      "loss": 0.1431,
+      "step": 16333
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012158156911581569,
+      "loss": 0.1383,
+      "step": 16334
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012156911581569116,
+      "loss": 0.2233,
+      "step": 16335
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012155666251556662,
+      "loss": 0.2868,
+      "step": 16336
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012154420921544208,
+      "loss": 0.1353,
+      "step": 16337
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012153175591531754,
+      "loss": 0.2825,
+      "step": 16338
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012151930261519301,
+      "loss": 0.1421,
+      "step": 16339
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012150684931506848,
+      "loss": 0.2662,
+      "step": 16340
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012149439601494394,
+      "loss": 0.1428,
+      "step": 16341
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012148194271481941,
+      "loss": 0.2733,
+      "step": 16342
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012146948941469489,
+      "loss": 0.1569,
+      "step": 16343
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012145703611457035,
+      "loss": 0.1485,
+      "step": 16344
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00012144458281444582,
+      "loss": 0.1752,
+      "step": 16345
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012143212951432129,
+      "loss": 0.1343,
+      "step": 16346
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012141967621419675,
+      "loss": 0.1594,
+      "step": 16347
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012140722291407222,
+      "loss": 0.3438,
+      "step": 16348
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001213947696139477,
+      "loss": 0.1622,
+      "step": 16349
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012138231631382315,
+      "loss": 0.1746,
+      "step": 16350
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012136986301369863,
+      "loss": 0.14,
+      "step": 16351
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001213574097135741,
+      "loss": 0.1118,
+      "step": 16352
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012134495641344955,
+      "loss": 0.2429,
+      "step": 16353
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012133250311332502,
+      "loss": 0.2512,
+      "step": 16354
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012132004981320048,
+      "loss": 0.1385,
+      "step": 16355
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012130759651307595,
+      "loss": 0.202,
+      "step": 16356
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012129514321295142,
+      "loss": 0.2078,
+      "step": 16357
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012128268991282688,
+      "loss": 0.1251,
+      "step": 16358
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012127023661270235,
+      "loss": 0.275,
+      "step": 16359
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012125778331257783,
+      "loss": 0.1402,
+      "step": 16360
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012124533001245329,
+      "loss": 0.1817,
+      "step": 16361
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012123287671232876,
+      "loss": 0.1469,
+      "step": 16362
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012122042341220423,
+      "loss": 0.2169,
+      "step": 16363
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012120797011207969,
+      "loss": 0.1193,
+      "step": 16364
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012119551681195516,
+      "loss": 0.2012,
+      "step": 16365
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012118306351183064,
+      "loss": 0.133,
+      "step": 16366
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0001211706102117061,
+      "loss": 0.1997,
+      "step": 16367
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012115815691158157,
+      "loss": 0.1748,
+      "step": 16368
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012114570361145701,
+      "loss": 0.1213,
+      "step": 16369
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012113325031133249,
+      "loss": 0.1537,
+      "step": 16370
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00012112079701120796,
+      "loss": 0.1729,
+      "step": 16371
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012110834371108342,
+      "loss": 0.1734,
+      "step": 16372
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012109589041095889,
+      "loss": 0.2425,
+      "step": 16373
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012108343711083436,
+      "loss": 0.1587,
+      "step": 16374
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012107098381070982,
+      "loss": 0.1601,
+      "step": 16375
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001210585305105853,
+      "loss": 0.1306,
+      "step": 16376
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012104607721046077,
+      "loss": 0.1563,
+      "step": 16377
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012103362391033623,
+      "loss": 0.1593,
+      "step": 16378
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001210211706102117,
+      "loss": 0.1758,
+      "step": 16379
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012100871731008716,
+      "loss": 0.2604,
+      "step": 16380
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012099626400996263,
+      "loss": 0.1943,
+      "step": 16381
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001209838107098381,
+      "loss": 0.1443,
+      "step": 16382
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012097135740971356,
+      "loss": 0.1629,
+      "step": 16383
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012095890410958904,
+      "loss": 0.1018,
+      "step": 16384
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001209464508094645,
+      "loss": 0.2286,
+      "step": 16385
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012093399750933996,
+      "loss": 0.2527,
+      "step": 16386
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012092154420921543,
+      "loss": 0.16,
+      "step": 16387
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001209090909090909,
+      "loss": 0.1857,
+      "step": 16388
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012089663760896636,
+      "loss": 0.1101,
+      "step": 16389
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012088418430884183,
+      "loss": 0.175,
+      "step": 16390
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012087173100871729,
+      "loss": 0.1177,
+      "step": 16391
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012085927770859276,
+      "loss": 0.1974,
+      "step": 16392
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012084682440846824,
+      "loss": 0.1699,
+      "step": 16393
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001208343711083437,
+      "loss": 0.1643,
+      "step": 16394
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012082191780821917,
+      "loss": 0.2581,
+      "step": 16395
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00012080946450809464,
+      "loss": 0.1547,
+      "step": 16396
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0001207970112079701,
+      "loss": 0.2552,
+      "step": 16397
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012078455790784557,
+      "loss": 0.2157,
+      "step": 16398
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012077210460772105,
+      "loss": 0.1691,
+      "step": 16399
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012075965130759649,
+      "loss": 0.1771,
+      "step": 16400
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012074719800747197,
+      "loss": 0.1592,
+      "step": 16401
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012073474470734742,
+      "loss": 0.1332,
+      "step": 16402
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001207222914072229,
+      "loss": 0.1468,
+      "step": 16403
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012070983810709837,
+      "loss": 0.1752,
+      "step": 16404
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012069738480697383,
+      "loss": 0.1904,
+      "step": 16405
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001206849315068493,
+      "loss": 0.1705,
+      "step": 16406
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012067247820672477,
+      "loss": 0.0781,
+      "step": 16407
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012066002490660023,
+      "loss": 0.1374,
+      "step": 16408
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001206475716064757,
+      "loss": 0.0945,
+      "step": 16409
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012063511830635118,
+      "loss": 0.1456,
+      "step": 16410
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012062266500622664,
+      "loss": 0.266,
+      "step": 16411
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012061021170610211,
+      "loss": 0.1696,
+      "step": 16412
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012059775840597758,
+      "loss": 0.1604,
+      "step": 16413
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012058530510585304,
+      "loss": 0.1268,
+      "step": 16414
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012057285180572852,
+      "loss": 0.1188,
+      "step": 16415
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012056039850560396,
+      "loss": 0.1334,
+      "step": 16416
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012054794520547943,
+      "loss": 0.2855,
+      "step": 16417
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0001205354919053549,
+      "loss": 0.2166,
+      "step": 16418
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012052303860523037,
+      "loss": 0.1424,
+      "step": 16419
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012051058530510584,
+      "loss": 0.121,
+      "step": 16420
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012049813200498131,
+      "loss": 0.1235,
+      "step": 16421
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012048567870485677,
+      "loss": 0.1358,
+      "step": 16422
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00012047322540473224,
+      "loss": 0.1789,
+      "step": 16423
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012046077210460772,
+      "loss": 0.1727,
+      "step": 16424
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012044831880448317,
+      "loss": 0.1389,
+      "step": 16425
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012043586550435865,
+      "loss": 0.2108,
+      "step": 16426
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012042341220423412,
+      "loss": 0.1685,
+      "step": 16427
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012041095890410958,
+      "loss": 0.1318,
+      "step": 16428
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012039850560398505,
+      "loss": 0.2425,
+      "step": 16429
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012038605230386052,
+      "loss": 0.1428,
+      "step": 16430
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012037359900373598,
+      "loss": 0.1532,
+      "step": 16431
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012036114570361144,
+      "loss": 0.1541,
+      "step": 16432
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0001203486924034869,
+      "loss": 0.2236,
+      "step": 16433
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012033623910336238,
+      "loss": 0.1389,
+      "step": 16434
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012032378580323785,
+      "loss": 0.1394,
+      "step": 16435
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012031133250311331,
+      "loss": 0.283,
+      "step": 16436
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012029887920298878,
+      "loss": 0.2131,
+      "step": 16437
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012028642590286425,
+      "loss": 0.2012,
+      "step": 16438
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012027397260273971,
+      "loss": 0.1362,
+      "step": 16439
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012026151930261518,
+      "loss": 0.1704,
+      "step": 16440
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012024906600249066,
+      "loss": 0.1609,
+      "step": 16441
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012023661270236612,
+      "loss": 0.2924,
+      "step": 16442
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012022415940224159,
+      "loss": 0.172,
+      "step": 16443
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012021170610211706,
+      "loss": 0.111,
+      "step": 16444
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012019925280199252,
+      "loss": 0.1881,
+      "step": 16445
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.000120186799501868,
+      "loss": 0.1606,
+      "step": 16446
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012017434620174344,
+      "loss": 0.1579,
+      "step": 16447
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012016189290161891,
+      "loss": 0.2148,
+      "step": 16448
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00012014943960149438,
+      "loss": 0.1688,
+      "step": 16449
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00012013698630136984,
+      "loss": 0.0949,
+      "step": 16450
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00012012453300124532,
+      "loss": 0.1792,
+      "step": 16451
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00012011207970112079,
+      "loss": 0.143,
+      "step": 16452
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00012009962640099625,
+      "loss": 0.1912,
+      "step": 16453
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00012008717310087172,
+      "loss": 0.2036,
+      "step": 16454
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0001200747198007472,
+      "loss": 0.1718,
+      "step": 16455
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00012006226650062265,
+      "loss": 0.3037,
+      "step": 16456
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00012004981320049813,
+      "loss": 0.2143,
+      "step": 16457
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0001200373599003736,
+      "loss": 0.1326,
+      "step": 16458
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00012002490660024906,
+      "loss": 0.1124,
+      "step": 16459
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00012001245330012453,
+      "loss": 0.1333,
+      "step": 16460
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 0.2681,
+      "step": 16461
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011998754669987546,
+      "loss": 0.1876,
+      "step": 16462
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011997509339975092,
+      "loss": 0.1407,
+      "step": 16463
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011996264009962638,
+      "loss": 0.1103,
+      "step": 16464
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011995018679950185,
+      "loss": 0.1145,
+      "step": 16465
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011993773349937733,
+      "loss": 0.1486,
+      "step": 16466
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011992528019925279,
+      "loss": 0.2184,
+      "step": 16467
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011991282689912826,
+      "loss": 0.1999,
+      "step": 16468
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011990037359900373,
+      "loss": 0.2084,
+      "step": 16469
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011988792029887919,
+      "loss": 0.1144,
+      "step": 16470
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011987546699875466,
+      "loss": 0.1329,
+      "step": 16471
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011986301369863012,
+      "loss": 0.0956,
+      "step": 16472
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0001198505603985056,
+      "loss": 0.2504,
+      "step": 16473
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011983810709838107,
+      "loss": 0.1998,
+      "step": 16474
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011982565379825653,
+      "loss": 0.223,
+      "step": 16475
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.000119813200498132,
+      "loss": 0.1989,
+      "step": 16476
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011980074719800747,
+      "loss": 0.1131,
+      "step": 16477
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011978829389788293,
+      "loss": 0.185,
+      "step": 16478
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011977584059775839,
+      "loss": 0.1615,
+      "step": 16479
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011976338729763385,
+      "loss": 0.2981,
+      "step": 16480
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011975093399750932,
+      "loss": 0.176,
+      "step": 16481
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001197384806973848,
+      "loss": 0.1569,
+      "step": 16482
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011972602739726025,
+      "loss": 0.2113,
+      "step": 16483
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011971357409713573,
+      "loss": 0.1178,
+      "step": 16484
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001197011207970112,
+      "loss": 0.291,
+      "step": 16485
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011968866749688666,
+      "loss": 0.2602,
+      "step": 16486
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011967621419676213,
+      "loss": 0.1942,
+      "step": 16487
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001196637608966376,
+      "loss": 0.1416,
+      "step": 16488
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011965130759651306,
+      "loss": 0.1131,
+      "step": 16489
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011963885429638854,
+      "loss": 0.2248,
+      "step": 16490
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011962640099626401,
+      "loss": 0.1617,
+      "step": 16491
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011961394769613947,
+      "loss": 0.2829,
+      "step": 16492
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011960149439601494,
+      "loss": 0.178,
+      "step": 16493
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011958904109589039,
+      "loss": 0.2344,
+      "step": 16494
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011957658779576586,
+      "loss": 0.1709,
+      "step": 16495
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011956413449564133,
+      "loss": 0.1043,
+      "step": 16496
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011955168119551679,
+      "loss": 0.1418,
+      "step": 16497
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011953922789539226,
+      "loss": 0.3225,
+      "step": 16498
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011952677459526774,
+      "loss": 0.1404,
+      "step": 16499
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0001195143212951432,
+      "loss": 0.1859,
+      "step": 16500
+    },
+    {
+      "epoch": 6.32,
+      "eval_cer": 0.10334198255847982,
+      "eval_loss": 0.3903411626815796,
+      "eval_runtime": 46.496,
+      "eval_samples_per_second": 37.702,
+      "eval_steps_per_second": 4.732,
+      "eval_wer": 0.16669683667466173,
+      "step": 16500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011950186799501867,
+      "loss": 0.1445,
+      "step": 16501
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011948941469489414,
+      "loss": 0.1504,
+      "step": 16502
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001194769613947696,
+      "loss": 0.1188,
+      "step": 16503
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011946450809464507,
+      "loss": 0.1947,
+      "step": 16504
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011945205479452055,
+      "loss": 0.1942,
+      "step": 16505
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.000119439601494396,
+      "loss": 0.1533,
+      "step": 16506
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011942714819427148,
+      "loss": 0.1818,
+      "step": 16507
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011941469489414695,
+      "loss": 0.1342,
+      "step": 16508
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011940224159402241,
+      "loss": 0.1909,
+      "step": 16509
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011938978829389787,
+      "loss": 0.1665,
+      "step": 16510
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011937733499377333,
+      "loss": 0.2165,
+      "step": 16511
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001193648816936488,
+      "loss": 0.148,
+      "step": 16512
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011935242839352427,
+      "loss": 0.1732,
+      "step": 16513
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011933997509339973,
+      "loss": 0.1028,
+      "step": 16514
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0001193275217932752,
+      "loss": 0.1402,
+      "step": 16515
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011931506849315068,
+      "loss": 0.1634,
+      "step": 16516
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011930261519302614,
+      "loss": 0.3349,
+      "step": 16517
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011929016189290161,
+      "loss": 0.1842,
+      "step": 16518
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011927770859277708,
+      "loss": 0.151,
+      "step": 16519
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011926525529265254,
+      "loss": 0.2112,
+      "step": 16520
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011925280199252801,
+      "loss": 0.1597,
+      "step": 16521
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011924034869240349,
+      "loss": 0.1913,
+      "step": 16522
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011922789539227895,
+      "loss": 0.2899,
+      "step": 16523
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011921544209215442,
+      "loss": 0.2085,
+      "step": 16524
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011920298879202989,
+      "loss": 0.1965,
+      "step": 16525
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011919053549190534,
+      "loss": 0.1288,
+      "step": 16526
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011917808219178081,
+      "loss": 0.1258,
+      "step": 16527
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011916562889165627,
+      "loss": 0.1776,
+      "step": 16528
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011915317559153174,
+      "loss": 0.3012,
+      "step": 16529
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011914072229140721,
+      "loss": 0.2943,
+      "step": 16530
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011912826899128267,
+      "loss": 0.284,
+      "step": 16531
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011911581569115815,
+      "loss": 0.1284,
+      "step": 16532
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011910336239103362,
+      "loss": 0.1069,
+      "step": 16533
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011909090909090908,
+      "loss": 0.1501,
+      "step": 16534
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011907845579078455,
+      "loss": 0.1444,
+      "step": 16535
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011906600249066002,
+      "loss": 0.2427,
+      "step": 16536
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011905354919053548,
+      "loss": 0.201,
+      "step": 16537
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011904109589041096,
+      "loss": 0.2807,
+      "step": 16538
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011902864259028643,
+      "loss": 0.1846,
+      "step": 16539
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011901618929016189,
+      "loss": 0.148,
+      "step": 16540
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011900373599003736,
+      "loss": 0.254,
+      "step": 16541
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0001189912826899128,
+      "loss": 0.2083,
+      "step": 16542
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011897882938978828,
+      "loss": 0.1305,
+      "step": 16543
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011896637608966375,
+      "loss": 0.1597,
+      "step": 16544
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011895392278953921,
+      "loss": 0.1144,
+      "step": 16545
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011894146948941468,
+      "loss": 0.1824,
+      "step": 16546
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011892901618929016,
+      "loss": 0.1913,
+      "step": 16547
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011891656288916561,
+      "loss": 0.417,
+      "step": 16548
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011890410958904109,
+      "loss": 0.1716,
+      "step": 16549
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011889165628891656,
+      "loss": 0.173,
+      "step": 16550
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011887920298879202,
+      "loss": 0.1595,
+      "step": 16551
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011886674968866749,
+      "loss": 0.1481,
+      "step": 16552
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011885429638854295,
+      "loss": 0.2393,
+      "step": 16553
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00011884184308841842,
+      "loss": 0.2761,
+      "step": 16554
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0001188293897882939,
+      "loss": 0.1691,
+      "step": 16555
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011881693648816936,
+      "loss": 0.2008,
+      "step": 16556
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011880448318804482,
+      "loss": 0.1614,
+      "step": 16557
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011879202988792029,
+      "loss": 0.1847,
+      "step": 16558
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011877957658779575,
+      "loss": 0.1236,
+      "step": 16559
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011876712328767122,
+      "loss": 0.1721,
+      "step": 16560
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011875466998754668,
+      "loss": 0.1917,
+      "step": 16561
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011874221668742215,
+      "loss": 0.2868,
+      "step": 16562
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011872976338729762,
+      "loss": 0.0955,
+      "step": 16563
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011871731008717308,
+      "loss": 0.1601,
+      "step": 16564
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011870485678704856,
+      "loss": 0.1388,
+      "step": 16565
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011869240348692403,
+      "loss": 0.2471,
+      "step": 16566
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011867995018679949,
+      "loss": 0.3062,
+      "step": 16567
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011866749688667496,
+      "loss": 0.2022,
+      "step": 16568
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011865504358655043,
+      "loss": 0.1575,
+      "step": 16569
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011864259028642589,
+      "loss": 0.1167,
+      "step": 16570
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011863013698630137,
+      "loss": 0.1824,
+      "step": 16571
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011861768368617684,
+      "loss": 0.1945,
+      "step": 16572
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011860523038605228,
+      "loss": 0.2187,
+      "step": 16573
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011859277708592776,
+      "loss": 0.2362,
+      "step": 16574
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011858032378580322,
+      "loss": 0.1861,
+      "step": 16575
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011856787048567869,
+      "loss": 0.2089,
+      "step": 16576
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011855541718555416,
+      "loss": 0.188,
+      "step": 16577
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011854296388542962,
+      "loss": 0.1424,
+      "step": 16578
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011853051058530509,
+      "loss": 0.2588,
+      "step": 16579
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011851805728518057,
+      "loss": 0.2464,
+      "step": 16580
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011850560398505602,
+      "loss": 0.131,
+      "step": 16581
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0001184931506849315,
+      "loss": 0.1514,
+      "step": 16582
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011848069738480697,
+      "loss": 0.1417,
+      "step": 16583
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011846824408468243,
+      "loss": 0.1498,
+      "step": 16584
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0001184557907845579,
+      "loss": 0.2277,
+      "step": 16585
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011844333748443337,
+      "loss": 0.2291,
+      "step": 16586
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011843088418430883,
+      "loss": 0.1274,
+      "step": 16587
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011841843088418431,
+      "loss": 0.2534,
+      "step": 16588
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011840597758405975,
+      "loss": 0.1512,
+      "step": 16589
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011839352428393523,
+      "loss": 0.1354,
+      "step": 16590
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0001183810709838107,
+      "loss": 0.163,
+      "step": 16591
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011836861768368616,
+      "loss": 0.2367,
+      "step": 16592
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011835616438356163,
+      "loss": 0.185,
+      "step": 16593
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0001183437110834371,
+      "loss": 0.1453,
+      "step": 16594
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011833125778331256,
+      "loss": 0.2238,
+      "step": 16595
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011831880448318803,
+      "loss": 0.1302,
+      "step": 16596
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011830635118306351,
+      "loss": 0.1768,
+      "step": 16597
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011829389788293897,
+      "loss": 0.3313,
+      "step": 16598
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011828144458281444,
+      "loss": 0.1573,
+      "step": 16599
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011826899128268991,
+      "loss": 0.2118,
+      "step": 16600
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011825653798256537,
+      "loss": 0.1526,
+      "step": 16601
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011824408468244084,
+      "loss": 0.1985,
+      "step": 16602
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011823163138231632,
+      "loss": 0.1257,
+      "step": 16603
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011821917808219176,
+      "loss": 0.2357,
+      "step": 16604
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011820672478206723,
+      "loss": 0.1921,
+      "step": 16605
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0001181942714819427,
+      "loss": 0.113,
+      "step": 16606
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011818181818181817,
+      "loss": 0.2626,
+      "step": 16607
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011816936488169364,
+      "loss": 0.1538,
+      "step": 16608
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0001181569115815691,
+      "loss": 0.1403,
+      "step": 16609
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011814445828144457,
+      "loss": 0.1648,
+      "step": 16610
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011813200498132004,
+      "loss": 0.1905,
+      "step": 16611
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0001181195516811955,
+      "loss": 0.1327,
+      "step": 16612
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011810709838107098,
+      "loss": 0.1543,
+      "step": 16613
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011809464508094645,
+      "loss": 0.1355,
+      "step": 16614
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011808219178082191,
+      "loss": 0.2154,
+      "step": 16615
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011806973848069738,
+      "loss": 0.2058,
+      "step": 16616
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011805728518057285,
+      "loss": 0.218,
+      "step": 16617
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011804483188044831,
+      "loss": 0.1473,
+      "step": 16618
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011803237858032378,
+      "loss": 0.267,
+      "step": 16619
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011801992528019923,
+      "loss": 0.1629,
+      "step": 16620
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0001180074719800747,
+      "loss": 0.1198,
+      "step": 16621
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011799501867995018,
+      "loss": 0.1835,
+      "step": 16622
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011798256537982564,
+      "loss": 0.2428,
+      "step": 16623
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011797011207970111,
+      "loss": 0.2074,
+      "step": 16624
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011795765877957658,
+      "loss": 0.1265,
+      "step": 16625
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011794520547945204,
+      "loss": 0.1444,
+      "step": 16626
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011793275217932751,
+      "loss": 0.1356,
+      "step": 16627
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011792029887920299,
+      "loss": 0.1241,
+      "step": 16628
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011790784557907844,
+      "loss": 0.1531,
+      "step": 16629
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011789539227895392,
+      "loss": 0.1782,
+      "step": 16630
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011788293897882939,
+      "loss": 0.1725,
+      "step": 16631
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011787048567870485,
+      "loss": 0.1426,
+      "step": 16632
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011785803237858032,
+      "loss": 0.1173,
+      "step": 16633
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011784557907845578,
+      "loss": 0.124,
+      "step": 16634
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011783312577833125,
+      "loss": 0.2062,
+      "step": 16635
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011782067247820671,
+      "loss": 0.2544,
+      "step": 16636
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011780821917808217,
+      "loss": 0.1549,
+      "step": 16637
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011779576587795764,
+      "loss": 0.1992,
+      "step": 16638
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011778331257783312,
+      "loss": 0.2384,
+      "step": 16639
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011777085927770858,
+      "loss": 0.1979,
+      "step": 16640
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011775840597758405,
+      "loss": 0.2315,
+      "step": 16641
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011774595267745951,
+      "loss": 0.38,
+      "step": 16642
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011773349937733498,
+      "loss": 0.1948,
+      "step": 16643
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011772104607721045,
+      "loss": 0.1412,
+      "step": 16644
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011770859277708591,
+      "loss": 0.224,
+      "step": 16645
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011769613947696139,
+      "loss": 0.1096,
+      "step": 16646
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011768368617683686,
+      "loss": 0.169,
+      "step": 16647
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011767123287671232,
+      "loss": 0.3149,
+      "step": 16648
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011765877957658779,
+      "loss": 0.159,
+      "step": 16649
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011764632627646326,
+      "loss": 0.1956,
+      "step": 16650
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011763387297633872,
+      "loss": 0.1627,
+      "step": 16651
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011762141967621418,
+      "loss": 0.1223,
+      "step": 16652
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011760896637608964,
+      "loss": 0.1109,
+      "step": 16653
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011759651307596511,
+      "loss": 0.2563,
+      "step": 16654
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011758405977584059,
+      "loss": 0.2891,
+      "step": 16655
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011757160647571605,
+      "loss": 0.1718,
+      "step": 16656
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011755915317559152,
+      "loss": 0.199,
+      "step": 16657
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011754669987546699,
+      "loss": 0.1447,
+      "step": 16658
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011753424657534245,
+      "loss": 0.2815,
+      "step": 16659
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011752179327521792,
+      "loss": 0.2821,
+      "step": 16660
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001175093399750934,
+      "loss": 0.2432,
+      "step": 16661
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011749688667496885,
+      "loss": 0.2084,
+      "step": 16662
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011748443337484433,
+      "loss": 0.1362,
+      "step": 16663
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001174719800747198,
+      "loss": 0.1464,
+      "step": 16664
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011745952677459526,
+      "loss": 0.1561,
+      "step": 16665
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011744707347447073,
+      "loss": 0.1288,
+      "step": 16666
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011743462017434618,
+      "loss": 0.2264,
+      "step": 16667
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011742216687422165,
+      "loss": 0.2552,
+      "step": 16668
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011740971357409712,
+      "loss": 0.2436,
+      "step": 16669
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011739726027397258,
+      "loss": 0.1101,
+      "step": 16670
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011738480697384805,
+      "loss": 0.1476,
+      "step": 16671
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011737235367372353,
+      "loss": 0.2699,
+      "step": 16672
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011735990037359899,
+      "loss": 0.2148,
+      "step": 16673
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011734744707347446,
+      "loss": 0.238,
+      "step": 16674
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011733499377334993,
+      "loss": 0.2024,
+      "step": 16675
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011732254047322539,
+      "loss": 0.23,
+      "step": 16676
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011731008717310086,
+      "loss": 0.1129,
+      "step": 16677
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011729763387297634,
+      "loss": 0.1432,
+      "step": 16678
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001172851805728518,
+      "loss": 0.1909,
+      "step": 16679
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011727272727272727,
+      "loss": 0.2049,
+      "step": 16680
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011726027397260274,
+      "loss": 0.2501,
+      "step": 16681
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0001172478206724782,
+      "loss": 0.1965,
+      "step": 16682
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011723536737235366,
+      "loss": 0.1134,
+      "step": 16683
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.00011722291407222912,
+      "loss": 0.1491,
+      "step": 16684
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011721046077210459,
+      "loss": 0.1577,
+      "step": 16685
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011719800747198006,
+      "loss": 0.2499,
+      "step": 16686
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011718555417185552,
+      "loss": 0.2796,
+      "step": 16687
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.000117173100871731,
+      "loss": 0.1532,
+      "step": 16688
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011716064757160647,
+      "loss": 0.1638,
+      "step": 16689
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011714819427148193,
+      "loss": 0.1486,
+      "step": 16690
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0001171357409713574,
+      "loss": 0.1451,
+      "step": 16691
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011712328767123287,
+      "loss": 0.2232,
+      "step": 16692
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011711083437110833,
+      "loss": 0.1491,
+      "step": 16693
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0001170983810709838,
+      "loss": 0.2297,
+      "step": 16694
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011708592777085928,
+      "loss": 0.1304,
+      "step": 16695
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011707347447073474,
+      "loss": 0.1292,
+      "step": 16696
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011706102117061021,
+      "loss": 0.1245,
+      "step": 16697
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011704856787048568,
+      "loss": 0.2403,
+      "step": 16698
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011703611457036113,
+      "loss": 0.1902,
+      "step": 16699
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0001170236612702366,
+      "loss": 0.1402,
+      "step": 16700
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011701120797011206,
+      "loss": 0.1359,
+      "step": 16701
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011699875466998753,
+      "loss": 0.1809,
+      "step": 16702
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.000116986301369863,
+      "loss": 0.1845,
+      "step": 16703
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011697384806973846,
+      "loss": 0.2042,
+      "step": 16704
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011696139476961394,
+      "loss": 0.2282,
+      "step": 16705
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011694894146948941,
+      "loss": 0.2004,
+      "step": 16706
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011693648816936487,
+      "loss": 0.1838,
+      "step": 16707
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011692403486924034,
+      "loss": 0.2609,
+      "step": 16708
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011691158156911581,
+      "loss": 0.1483,
+      "step": 16709
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00011689912826899127,
+      "loss": 0.1343,
+      "step": 16710
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011688667496886675,
+      "loss": 0.2274,
+      "step": 16711
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011687422166874222,
+      "loss": 0.1706,
+      "step": 16712
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011686176836861768,
+      "loss": 0.1153,
+      "step": 16713
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011684931506849314,
+      "loss": 0.229,
+      "step": 16714
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001168368617683686,
+      "loss": 0.1321,
+      "step": 16715
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011682440846824407,
+      "loss": 0.1898,
+      "step": 16716
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011681195516811954,
+      "loss": 0.169,
+      "step": 16717
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.000116799501867995,
+      "loss": 0.3266,
+      "step": 16718
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011678704856787047,
+      "loss": 0.19,
+      "step": 16719
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011677459526774595,
+      "loss": 0.1266,
+      "step": 16720
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001167621419676214,
+      "loss": 0.1106,
+      "step": 16721
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011674968866749688,
+      "loss": 0.1231,
+      "step": 16722
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011673723536737234,
+      "loss": 0.3167,
+      "step": 16723
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011672478206724781,
+      "loss": 0.2699,
+      "step": 16724
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011671232876712328,
+      "loss": 0.1869,
+      "step": 16725
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011669987546699874,
+      "loss": 0.1521,
+      "step": 16726
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011668742216687422,
+      "loss": 0.1511,
+      "step": 16727
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011667496886674969,
+      "loss": 0.1323,
+      "step": 16728
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011666251556662515,
+      "loss": 0.3029,
+      "step": 16729
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0001166500622665006,
+      "loss": 0.2769,
+      "step": 16730
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011663760896637608,
+      "loss": 0.1476,
+      "step": 16731
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011662515566625154,
+      "loss": 0.176,
+      "step": 16732
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011661270236612701,
+      "loss": 0.1929,
+      "step": 16733
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011660024906600247,
+      "loss": 0.1332,
+      "step": 16734
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011658779576587794,
+      "loss": 0.2103,
+      "step": 16735
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011657534246575342,
+      "loss": 0.2816,
+      "step": 16736
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011656288916562887,
+      "loss": 0.1594,
+      "step": 16737
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011655043586550435,
+      "loss": 0.1635,
+      "step": 16738
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011653798256537982,
+      "loss": 0.1281,
+      "step": 16739
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011652552926525528,
+      "loss": 0.1442,
+      "step": 16740
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011651307596513075,
+      "loss": 0.1571,
+      "step": 16741
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011650062266500622,
+      "loss": 0.346,
+      "step": 16742
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011648816936488168,
+      "loss": 0.1351,
+      "step": 16743
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011647571606475716,
+      "loss": 0.1534,
+      "step": 16744
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011646326276463263,
+      "loss": 0.1184,
+      "step": 16745
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011645080946450808,
+      "loss": 0.1669,
+      "step": 16746
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011643835616438355,
+      "loss": 0.1219,
+      "step": 16747
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011642590286425901,
+      "loss": 0.2673,
+      "step": 16748
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011641344956413448,
+      "loss": 0.1625,
+      "step": 16749
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011640099626400995,
+      "loss": 0.2704,
+      "step": 16750
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011638854296388541,
+      "loss": 0.2464,
+      "step": 16751
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011637608966376088,
+      "loss": 0.1827,
+      "step": 16752
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011636363636363636,
+      "loss": 0.1029,
+      "step": 16753
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011635118306351182,
+      "loss": 0.2292,
+      "step": 16754
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011633872976338729,
+      "loss": 0.1785,
+      "step": 16755
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011632627646326276,
+      "loss": 0.1815,
+      "step": 16756
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011631382316313822,
+      "loss": 0.1816,
+      "step": 16757
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0001163013698630137,
+      "loss": 0.1204,
+      "step": 16758
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011628891656288917,
+      "loss": 0.2833,
+      "step": 16759
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011627646326276463,
+      "loss": 0.144,
+      "step": 16760
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011626400996264008,
+      "loss": 0.3001,
+      "step": 16761
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011625155666251554,
+      "loss": 0.3159,
+      "step": 16762
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011623910336239102,
+      "loss": 0.1516,
+      "step": 16763
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011622665006226649,
+      "loss": 0.1059,
+      "step": 16764
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011621419676214195,
+      "loss": 0.1795,
+      "step": 16765
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011620174346201742,
+      "loss": 0.1412,
+      "step": 16766
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001161892901618929,
+      "loss": 0.3461,
+      "step": 16767
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011617683686176835,
+      "loss": 0.2258,
+      "step": 16768
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011616438356164383,
+      "loss": 0.2333,
+      "step": 16769
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001161519302615193,
+      "loss": 0.2003,
+      "step": 16770
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011613947696139476,
+      "loss": 0.1985,
+      "step": 16771
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011612702366127023,
+      "loss": 0.1568,
+      "step": 16772
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001161145703611457,
+      "loss": 0.2143,
+      "step": 16773
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011610211706102116,
+      "loss": 0.1884,
+      "step": 16774
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011608966376089664,
+      "loss": 0.1409,
+      "step": 16775
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011607721046077211,
+      "loss": 0.17,
+      "step": 16776
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011606475716064755,
+      "loss": 0.2258,
+      "step": 16777
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011605230386052303,
+      "loss": 0.1672,
+      "step": 16778
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011603985056039849,
+      "loss": 0.213,
+      "step": 16779
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011602739726027396,
+      "loss": 0.2459,
+      "step": 16780
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011601494396014943,
+      "loss": 0.1622,
+      "step": 16781
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011600249066002489,
+      "loss": 0.2168,
+      "step": 16782
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011599003735990036,
+      "loss": 0.1733,
+      "step": 16783
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011597758405977584,
+      "loss": 0.208,
+      "step": 16784
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001159651307596513,
+      "loss": 0.1342,
+      "step": 16785
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011595267745952677,
+      "loss": 0.2222,
+      "step": 16786
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.00011594022415940224,
+      "loss": 0.1332,
+      "step": 16787
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0001159277708592777,
+      "loss": 0.2943,
+      "step": 16788
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011591531755915317,
+      "loss": 0.2029,
+      "step": 16789
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011590286425902864,
+      "loss": 0.1545,
+      "step": 16790
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001158904109589041,
+      "loss": 0.1202,
+      "step": 16791
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011587795765877958,
+      "loss": 0.3462,
+      "step": 16792
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011586550435865502,
+      "loss": 0.2258,
+      "step": 16793
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001158530510585305,
+      "loss": 0.104,
+      "step": 16794
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011584059775840597,
+      "loss": 0.1351,
+      "step": 16795
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011582814445828143,
+      "loss": 0.1585,
+      "step": 16796
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001158156911581569,
+      "loss": 0.1677,
+      "step": 16797
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011580323785803237,
+      "loss": 0.2723,
+      "step": 16798
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011579078455790783,
+      "loss": 0.1755,
+      "step": 16799
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001157783312577833,
+      "loss": 0.1665,
+      "step": 16800
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011576587795765878,
+      "loss": 0.1457,
+      "step": 16801
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011575342465753424,
+      "loss": 0.1055,
+      "step": 16802
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011574097135740971,
+      "loss": 0.1785,
+      "step": 16803
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011572851805728517,
+      "loss": 0.1944,
+      "step": 16804
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011571606475716064,
+      "loss": 0.2136,
+      "step": 16805
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011570361145703611,
+      "loss": 0.2192,
+      "step": 16806
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011569115815691157,
+      "loss": 0.1692,
+      "step": 16807
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011567870485678705,
+      "loss": 0.0736,
+      "step": 16808
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001156662515566625,
+      "loss": 0.1333,
+      "step": 16809
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011565379825653796,
+      "loss": 0.2581,
+      "step": 16810
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011564134495641344,
+      "loss": 0.2113,
+      "step": 16811
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011562889165628891,
+      "loss": 0.1456,
+      "step": 16812
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011561643835616437,
+      "loss": 0.1571,
+      "step": 16813
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011560398505603984,
+      "loss": 0.14,
+      "step": 16814
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0001155915317559153,
+      "loss": 0.1257,
+      "step": 16815
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011557907845579077,
+      "loss": 0.3031,
+      "step": 16816
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011556662515566625,
+      "loss": 0.3227,
+      "step": 16817
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0001155541718555417,
+      "loss": 0.1439,
+      "step": 16818
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011554171855541718,
+      "loss": 0.133,
+      "step": 16819
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011552926525529265,
+      "loss": 0.187,
+      "step": 16820
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011551681195516811,
+      "loss": 0.2208,
+      "step": 16821
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011550435865504358,
+      "loss": 0.1602,
+      "step": 16822
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011549190535491905,
+      "loss": 0.3068,
+      "step": 16823
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0001154794520547945,
+      "loss": 0.1942,
+      "step": 16824
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011546699875466997,
+      "loss": 0.1225,
+      "step": 16825
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011545454545454543,
+      "loss": 0.1021,
+      "step": 16826
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0001154420921544209,
+      "loss": 0.2563,
+      "step": 16827
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011542963885429638,
+      "loss": 0.2726,
+      "step": 16828
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011541718555417184,
+      "loss": 0.2332,
+      "step": 16829
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011540473225404731,
+      "loss": 0.1806,
+      "step": 16830
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011539227895392278,
+      "loss": 0.144,
+      "step": 16831
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011537982565379824,
+      "loss": 0.1498,
+      "step": 16832
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011536737235367371,
+      "loss": 0.2041,
+      "step": 16833
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011535491905354919,
+      "loss": 0.1469,
+      "step": 16834
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011534246575342465,
+      "loss": 0.1667,
+      "step": 16835
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011533001245330012,
+      "loss": 0.2019,
+      "step": 16836
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011531755915317559,
+      "loss": 0.1604,
+      "step": 16837
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011530510585305105,
+      "loss": 0.2884,
+      "step": 16838
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011529265255292652,
+      "loss": 0.193,
+      "step": 16839
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011528019925280197,
+      "loss": 0.139,
+      "step": 16840
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011526774595267744,
+      "loss": 0.2082,
+      "step": 16841
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011525529265255291,
+      "loss": 0.239,
+      "step": 16842
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011524283935242837,
+      "loss": 0.188,
+      "step": 16843
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011523038605230385,
+      "loss": 0.1159,
+      "step": 16844
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011521793275217932,
+      "loss": 0.1714,
+      "step": 16845
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011520547945205478,
+      "loss": 0.2171,
+      "step": 16846
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011519302615193025,
+      "loss": 0.2489,
+      "step": 16847
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011518057285180572,
+      "loss": 0.3445,
+      "step": 16848
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011516811955168118,
+      "loss": 0.1308,
+      "step": 16849
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011515566625155666,
+      "loss": 0.1657,
+      "step": 16850
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011514321295143213,
+      "loss": 0.1321,
+      "step": 16851
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011513075965130759,
+      "loss": 0.111,
+      "step": 16852
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011511830635118306,
+      "loss": 0.146,
+      "step": 16853
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011510585305105853,
+      "loss": 0.2052,
+      "step": 16854
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011509339975093399,
+      "loss": 0.2501,
+      "step": 16855
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011508094645080945,
+      "loss": 0.1538,
+      "step": 16856
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011506849315068491,
+      "loss": 0.166,
+      "step": 16857
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011505603985056038,
+      "loss": 0.1213,
+      "step": 16858
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011504358655043586,
+      "loss": 0.1498,
+      "step": 16859
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011503113325031131,
+      "loss": 0.2,
+      "step": 16860
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011501867995018679,
+      "loss": 0.1857,
+      "step": 16861
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011500622665006226,
+      "loss": 0.2429,
+      "step": 16862
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011499377334993772,
+      "loss": 0.2245,
+      "step": 16863
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011498132004981319,
+      "loss": 0.0946,
+      "step": 16864
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011496886674968866,
+      "loss": 0.1785,
+      "step": 16865
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011495641344956412,
+      "loss": 0.1279,
+      "step": 16866
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0001149439601494396,
+      "loss": 0.2718,
+      "step": 16867
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011493150684931507,
+      "loss": 0.173,
+      "step": 16868
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011491905354919053,
+      "loss": 0.1979,
+      "step": 16869
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.000114906600249066,
+      "loss": 0.1548,
+      "step": 16870
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011489414694894145,
+      "loss": 0.1221,
+      "step": 16871
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011488169364881692,
+      "loss": 0.1186,
+      "step": 16872
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011486924034869239,
+      "loss": 0.2899,
+      "step": 16873
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011485678704856785,
+      "loss": 0.1325,
+      "step": 16874
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011484433374844332,
+      "loss": 0.1606,
+      "step": 16875
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0001148318804483188,
+      "loss": 0.181,
+      "step": 16876
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011481942714819426,
+      "loss": 0.1747,
+      "step": 16877
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011480697384806973,
+      "loss": 0.0981,
+      "step": 16878
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0001147945205479452,
+      "loss": 0.4593,
+      "step": 16879
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011478206724782066,
+      "loss": 0.1995,
+      "step": 16880
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011476961394769613,
+      "loss": 0.1962,
+      "step": 16881
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0001147571606475716,
+      "loss": 0.2363,
+      "step": 16882
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011474470734744707,
+      "loss": 0.1747,
+      "step": 16883
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011473225404732254,
+      "loss": 0.1629,
+      "step": 16884
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.000114719800747198,
+      "loss": 0.1391,
+      "step": 16885
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011470734744707347,
+      "loss": 0.2052,
+      "step": 16886
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011469489414694893,
+      "loss": 0.2029,
+      "step": 16887
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011468244084682439,
+      "loss": 0.2126,
+      "step": 16888
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011466998754669986,
+      "loss": 0.195,
+      "step": 16889
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011465753424657533,
+      "loss": 0.1709,
+      "step": 16890
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011464508094645079,
+      "loss": 0.2082,
+      "step": 16891
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011463262764632627,
+      "loss": 0.2673,
+      "step": 16892
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011462017434620174,
+      "loss": 0.1901,
+      "step": 16893
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0001146077210460772,
+      "loss": 0.2239,
+      "step": 16894
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011459526774595267,
+      "loss": 0.179,
+      "step": 16895
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011458281444582813,
+      "loss": 0.1315,
+      "step": 16896
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0001145703611457036,
+      "loss": 0.1384,
+      "step": 16897
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011455790784557908,
+      "loss": 0.307,
+      "step": 16898
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011454545454545453,
+      "loss": 0.1522,
+      "step": 16899
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011453300124533001,
+      "loss": 0.1672,
+      "step": 16900
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011452054794520548,
+      "loss": 0.2634,
+      "step": 16901
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011450809464508094,
+      "loss": 0.134,
+      "step": 16902
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0001144956413449564,
+      "loss": 0.3082,
+      "step": 16903
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011448318804483186,
+      "loss": 0.2297,
+      "step": 16904
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011447073474470733,
+      "loss": 0.1975,
+      "step": 16905
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0001144582814445828,
+      "loss": 0.1174,
+      "step": 16906
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011444582814445826,
+      "loss": 0.137,
+      "step": 16907
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011443337484433373,
+      "loss": 0.1388,
+      "step": 16908
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011442092154420921,
+      "loss": 0.121,
+      "step": 16909
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011440846824408467,
+      "loss": 0.1756,
+      "step": 16910
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011439601494396014,
+      "loss": 0.1742,
+      "step": 16911
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011438356164383561,
+      "loss": 0.1703,
+      "step": 16912
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011437110834371107,
+      "loss": 0.1799,
+      "step": 16913
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011435865504358654,
+      "loss": 0.1558,
+      "step": 16914
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011434620174346202,
+      "loss": 0.142,
+      "step": 16915
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011433374844333748,
+      "loss": 0.2123,
+      "step": 16916
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011432129514321295,
+      "loss": 0.312,
+      "step": 16917
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0001143088418430884,
+      "loss": 0.1916,
+      "step": 16918
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00011429638854296387,
+      "loss": 0.1457,
+      "step": 16919
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011428393524283934,
+      "loss": 0.1695,
+      "step": 16920
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001142714819427148,
+      "loss": 0.1666,
+      "step": 16921
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011425902864259027,
+      "loss": 0.1499,
+      "step": 16922
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011424657534246574,
+      "loss": 0.2558,
+      "step": 16923
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0001142341220423412,
+      "loss": 0.1777,
+      "step": 16924
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011422166874221668,
+      "loss": 0.1674,
+      "step": 16925
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011420921544209215,
+      "loss": 0.1549,
+      "step": 16926
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011419676214196761,
+      "loss": 0.1054,
+      "step": 16927
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011418430884184308,
+      "loss": 0.1509,
+      "step": 16928
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011417185554171855,
+      "loss": 0.1662,
+      "step": 16929
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011415940224159401,
+      "loss": 0.2186,
+      "step": 16930
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011414694894146949,
+      "loss": 0.2978,
+      "step": 16931
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011413449564134496,
+      "loss": 0.1281,
+      "step": 16932
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011412204234122042,
+      "loss": 0.0797,
+      "step": 16933
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011410958904109588,
+      "loss": 0.1719,
+      "step": 16934
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011409713574097134,
+      "loss": 0.1274,
+      "step": 16935
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011408468244084681,
+      "loss": 0.2751,
+      "step": 16936
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011407222914072228,
+      "loss": 0.1773,
+      "step": 16937
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011405977584059774,
+      "loss": 0.2253,
+      "step": 16938
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011404732254047321,
+      "loss": 0.1719,
+      "step": 16939
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011403486924034869,
+      "loss": 0.1982,
+      "step": 16940
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011402241594022414,
+      "loss": 0.1799,
+      "step": 16941
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011400996264009962,
+      "loss": 0.1909,
+      "step": 16942
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011399750933997509,
+      "loss": 0.1607,
+      "step": 16943
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011398505603985055,
+      "loss": 0.1209,
+      "step": 16944
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011397260273972602,
+      "loss": 0.2353,
+      "step": 16945
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001139601494396015,
+      "loss": 0.1807,
+      "step": 16946
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011394769613947695,
+      "loss": 0.2825,
+      "step": 16947
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011393524283935243,
+      "loss": 0.2339,
+      "step": 16948
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001139227895392279,
+      "loss": 0.1956,
+      "step": 16949
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011391033623910334,
+      "loss": 0.1438,
+      "step": 16950
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011389788293897882,
+      "loss": 0.0993,
+      "step": 16951
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011388542963885428,
+      "loss": 0.1133,
+      "step": 16952
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011387297633872975,
+      "loss": 0.141,
+      "step": 16953
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011386052303860522,
+      "loss": 0.2518,
+      "step": 16954
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011384806973848068,
+      "loss": 0.193,
+      "step": 16955
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011383561643835615,
+      "loss": 0.1313,
+      "step": 16956
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011382316313823163,
+      "loss": 0.1799,
+      "step": 16957
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011381070983810709,
+      "loss": 0.1321,
+      "step": 16958
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011379825653798256,
+      "loss": 0.1006,
+      "step": 16959
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011378580323785803,
+      "loss": 0.1681,
+      "step": 16960
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011377334993773349,
+      "loss": 0.3566,
+      "step": 16961
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011376089663760896,
+      "loss": 0.2614,
+      "step": 16962
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011374844333748444,
+      "loss": 0.158,
+      "step": 16963
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0001137359900373599,
+      "loss": 0.1146,
+      "step": 16964
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011372353673723537,
+      "loss": 0.178,
+      "step": 16965
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011371108343711081,
+      "loss": 0.1132,
+      "step": 16966
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011369863013698629,
+      "loss": 0.2123,
+      "step": 16967
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011368617683686176,
+      "loss": 0.2288,
+      "step": 16968
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011367372353673722,
+      "loss": 0.164,
+      "step": 16969
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011366127023661269,
+      "loss": 0.1914,
+      "step": 16970
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011364881693648816,
+      "loss": 0.1161,
+      "step": 16971
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011363636363636362,
+      "loss": 0.2467,
+      "step": 16972
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001136239103362391,
+      "loss": 0.2455,
+      "step": 16973
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011361145703611457,
+      "loss": 0.2152,
+      "step": 16974
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011359900373599003,
+      "loss": 0.1833,
+      "step": 16975
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001135865504358655,
+      "loss": 0.2366,
+      "step": 16976
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011357409713574096,
+      "loss": 0.1933,
+      "step": 16977
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011356164383561643,
+      "loss": 0.2196,
+      "step": 16978
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001135491905354919,
+      "loss": 0.295,
+      "step": 16979
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011353673723536736,
+      "loss": 0.2645,
+      "step": 16980
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011352428393524282,
+      "loss": 0.1777,
+      "step": 16981
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001135118306351183,
+      "loss": 0.1462,
+      "step": 16982
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011349937733499375,
+      "loss": 0.1738,
+      "step": 16983
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011348692403486923,
+      "loss": 0.1174,
+      "step": 16984
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011347447073474469,
+      "loss": 0.2722,
+      "step": 16985
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011346201743462016,
+      "loss": 0.2962,
+      "step": 16986
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011344956413449563,
+      "loss": 0.3462,
+      "step": 16987
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011343711083437109,
+      "loss": 0.2126,
+      "step": 16988
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011342465753424656,
+      "loss": 0.1564,
+      "step": 16989
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011341220423412204,
+      "loss": 0.2333,
+      "step": 16990
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001133997509339975,
+      "loss": 0.1786,
+      "step": 16991
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001133997509339975,
+      "loss": 0.7008,
+      "step": 16992
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011338729763387297,
+      "loss": 0.22,
+      "step": 16993
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011337484433374844,
+      "loss": 0.1788,
+      "step": 16994
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0001133623910336239,
+      "loss": 0.2,
+      "step": 16995
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011334993773349937,
+      "loss": 0.1358,
+      "step": 16996
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011333748443337485,
+      "loss": 0.1791,
+      "step": 16997
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011332503113325029,
+      "loss": 0.2968,
+      "step": 16998
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011331257783312576,
+      "loss": 0.2396,
+      "step": 16999
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011330012453300122,
+      "loss": 0.1374,
+      "step": 17000
+    },
+    {
+      "epoch": 6.52,
+      "eval_cer": 0.10480519733890005,
+      "eval_loss": 0.37208250164985657,
+      "eval_runtime": 47.5737,
+      "eval_samples_per_second": 36.848,
+      "eval_steps_per_second": 4.624,
+      "eval_wer": 0.1735982260035299,
+      "step": 17000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0001132876712328767,
+      "loss": 0.2168,
+      "step": 17001
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011327521793275217,
+      "loss": 0.1411,
+      "step": 17002
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011326276463262763,
+      "loss": 0.1655,
+      "step": 17003
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0001132503113325031,
+      "loss": 0.3336,
+      "step": 17004
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011323785803237857,
+      "loss": 0.2751,
+      "step": 17005
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011322540473225403,
+      "loss": 0.2933,
+      "step": 17006
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0001132129514321295,
+      "loss": 0.1435,
+      "step": 17007
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011320049813200498,
+      "loss": 0.223,
+      "step": 17008
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011318804483188044,
+      "loss": 0.2178,
+      "step": 17009
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011317559153175591,
+      "loss": 0.1104,
+      "step": 17010
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011316313823163138,
+      "loss": 0.2733,
+      "step": 17011
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011315068493150684,
+      "loss": 0.2134,
+      "step": 17012
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011313823163138231,
+      "loss": 0.2285,
+      "step": 17013
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011312577833125776,
+      "loss": 0.1333,
+      "step": 17014
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011311332503113323,
+      "loss": 0.1588,
+      "step": 17015
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0001131008717310087,
+      "loss": 0.1218,
+      "step": 17016
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011308841843088416,
+      "loss": 0.2906,
+      "step": 17017
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011307596513075964,
+      "loss": 0.169,
+      "step": 17018
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011306351183063511,
+      "loss": 0.22,
+      "step": 17019
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011305105853051057,
+      "loss": 0.1292,
+      "step": 17020
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011303860523038604,
+      "loss": 0.1735,
+      "step": 17021
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011302615193026152,
+      "loss": 0.1394,
+      "step": 17022
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00011301369863013697,
+      "loss": 0.3416,
+      "step": 17023
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011300124533001245,
+      "loss": 0.1591,
+      "step": 17024
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011298879202988792,
+      "loss": 0.3785,
+      "step": 17025
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011297633872976338,
+      "loss": 0.1652,
+      "step": 17026
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011296388542963885,
+      "loss": 0.1328,
+      "step": 17027
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011295143212951432,
+      "loss": 0.1212,
+      "step": 17028
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011293897882938977,
+      "loss": 0.3344,
+      "step": 17029
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011292652552926524,
+      "loss": 0.1618,
+      "step": 17030
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001129140722291407,
+      "loss": 0.1439,
+      "step": 17031
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011290161892901617,
+      "loss": 0.1223,
+      "step": 17032
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011288916562889165,
+      "loss": 0.1009,
+      "step": 17033
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0001128767123287671,
+      "loss": 0.2002,
+      "step": 17034
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011286425902864258,
+      "loss": 0.16,
+      "step": 17035
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011285180572851805,
+      "loss": 0.2294,
+      "step": 17036
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011283935242839351,
+      "loss": 0.27,
+      "step": 17037
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011282689912826898,
+      "loss": 0.1899,
+      "step": 17038
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011281444582814446,
+      "loss": 0.1247,
+      "step": 17039
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011280199252801992,
+      "loss": 0.0981,
+      "step": 17040
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011278953922789539,
+      "loss": 0.0829,
+      "step": 17041
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011277708592777086,
+      "loss": 0.1955,
+      "step": 17042
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011276463262764632,
+      "loss": 0.2886,
+      "step": 17043
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011275217932752179,
+      "loss": 0.1468,
+      "step": 17044
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011273972602739724,
+      "loss": 0.0886,
+      "step": 17045
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011272727272727271,
+      "loss": 0.1411,
+      "step": 17046
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011271481942714818,
+      "loss": 0.2531,
+      "step": 17047
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011270236612702364,
+      "loss": 0.2108,
+      "step": 17048
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.00011268991282689912,
+      "loss": 0.1848,
+      "step": 17049
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011267745952677459,
+      "loss": 0.1848,
+      "step": 17050
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011266500622665005,
+      "loss": 0.1773,
+      "step": 17051
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011265255292652552,
+      "loss": 0.1429,
+      "step": 17052
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011264009962640099,
+      "loss": 0.111,
+      "step": 17053
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011262764632627645,
+      "loss": 0.1396,
+      "step": 17054
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011261519302615193,
+      "loss": 0.1632,
+      "step": 17055
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001126027397260274,
+      "loss": 0.157,
+      "step": 17056
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011259028642590286,
+      "loss": 0.2239,
+      "step": 17057
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011257783312577833,
+      "loss": 0.1001,
+      "step": 17058
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011256537982565379,
+      "loss": 0.2061,
+      "step": 17059
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011255292652552926,
+      "loss": 0.1783,
+      "step": 17060
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011254047322540472,
+      "loss": 0.1827,
+      "step": 17061
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011252801992528018,
+      "loss": 0.2105,
+      "step": 17062
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011251556662515565,
+      "loss": 0.2258,
+      "step": 17063
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011250311332503113,
+      "loss": 0.1856,
+      "step": 17064
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011249066002490658,
+      "loss": 0.1209,
+      "step": 17065
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011247820672478206,
+      "loss": 0.1067,
+      "step": 17066
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011246575342465752,
+      "loss": 0.3957,
+      "step": 17067
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011245330012453299,
+      "loss": 0.1292,
+      "step": 17068
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011244084682440846,
+      "loss": 0.124,
+      "step": 17069
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011242839352428392,
+      "loss": 0.1525,
+      "step": 17070
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001124159402241594,
+      "loss": 0.1115,
+      "step": 17071
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011240348692403487,
+      "loss": 0.2225,
+      "step": 17072
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011239103362391033,
+      "loss": 0.247,
+      "step": 17073
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0001123785803237858,
+      "loss": 0.2077,
+      "step": 17074
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011236612702366127,
+      "loss": 0.1412,
+      "step": 17075
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011235367372353673,
+      "loss": 0.2278,
+      "step": 17076
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011234122042341219,
+      "loss": 0.146,
+      "step": 17077
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011232876712328765,
+      "loss": 0.1188,
+      "step": 17078
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011231631382316312,
+      "loss": 0.1945,
+      "step": 17079
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001123038605230386,
+      "loss": 0.2152,
+      "step": 17080
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011229140722291405,
+      "loss": 0.2748,
+      "step": 17081
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011227895392278953,
+      "loss": 0.1522,
+      "step": 17082
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.000112266500622665,
+      "loss": 0.1162,
+      "step": 17083
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011225404732254046,
+      "loss": 0.1172,
+      "step": 17084
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011224159402241593,
+      "loss": 0.1926,
+      "step": 17085
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001122291407222914,
+      "loss": 0.2742,
+      "step": 17086
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011221668742216686,
+      "loss": 0.2067,
+      "step": 17087
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011220423412204234,
+      "loss": 0.1644,
+      "step": 17088
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011219178082191781,
+      "loss": 0.1425,
+      "step": 17089
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011217932752179327,
+      "loss": 0.1071,
+      "step": 17090
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011216687422166874,
+      "loss": 0.1222,
+      "step": 17091
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011215442092154419,
+      "loss": 0.2576,
+      "step": 17092
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011214196762141966,
+      "loss": 0.2121,
+      "step": 17093
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011212951432129513,
+      "loss": 0.1328,
+      "step": 17094
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011211706102117059,
+      "loss": 0.1117,
+      "step": 17095
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011210460772104606,
+      "loss": 0.1276,
+      "step": 17096
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011209215442092154,
+      "loss": 0.2454,
+      "step": 17097
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.000112079701120797,
+      "loss": 0.2444,
+      "step": 17098
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011206724782067247,
+      "loss": 0.1322,
+      "step": 17099
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011205479452054794,
+      "loss": 0.1456,
+      "step": 17100
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0001120423412204234,
+      "loss": 0.2056,
+      "step": 17101
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011202988792029887,
+      "loss": 0.2071,
+      "step": 17102
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011201743462017434,
+      "loss": 0.1929,
+      "step": 17103
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0001120049813200498,
+      "loss": 0.2346,
+      "step": 17104
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011199252801992528,
+      "loss": 0.1597,
+      "step": 17105
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011198007471980075,
+      "loss": 0.1959,
+      "step": 17106
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011196762141967621,
+      "loss": 0.1746,
+      "step": 17107
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011195516811955167,
+      "loss": 0.1778,
+      "step": 17108
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011194271481942713,
+      "loss": 0.1726,
+      "step": 17109
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0001119302615193026,
+      "loss": 0.1653,
+      "step": 17110
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011191780821917807,
+      "loss": 0.3663,
+      "step": 17111
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011190535491905353,
+      "loss": 0.1714,
+      "step": 17112
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.000111892901618929,
+      "loss": 0.1749,
+      "step": 17113
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011188044831880448,
+      "loss": 0.1347,
+      "step": 17114
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011186799501867994,
+      "loss": 0.1205,
+      "step": 17115
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011185554171855541,
+      "loss": 0.1854,
+      "step": 17116
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011184308841843088,
+      "loss": 0.3843,
+      "step": 17117
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011183063511830634,
+      "loss": 0.2012,
+      "step": 17118
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011181818181818181,
+      "loss": 0.2498,
+      "step": 17119
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011180572851805729,
+      "loss": 0.1347,
+      "step": 17120
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011179327521793275,
+      "loss": 0.1276,
+      "step": 17121
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011178082191780822,
+      "loss": 0.1951,
+      "step": 17122
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011176836861768369,
+      "loss": 0.3586,
+      "step": 17123
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011175591531755914,
+      "loss": 0.3462,
+      "step": 17124
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011174346201743461,
+      "loss": 0.2156,
+      "step": 17125
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011173100871731007,
+      "loss": 0.1513,
+      "step": 17126
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011171855541718554,
+      "loss": 0.1279,
+      "step": 17127
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011170610211706101,
+      "loss": 0.099,
+      "step": 17128
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011169364881693647,
+      "loss": 0.1841,
+      "step": 17129
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011168119551681195,
+      "loss": 0.1917,
+      "step": 17130
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011166874221668742,
+      "loss": 0.2294,
+      "step": 17131
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011165628891656288,
+      "loss": 0.2226,
+      "step": 17132
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011164383561643835,
+      "loss": 0.1828,
+      "step": 17133
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011163138231631382,
+      "loss": 0.1605,
+      "step": 17134
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011161892901618928,
+      "loss": 0.1212,
+      "step": 17135
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011160647571606475,
+      "loss": 0.2278,
+      "step": 17136
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011159402241594023,
+      "loss": 0.1945,
+      "step": 17137
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011158156911581569,
+      "loss": 0.1253,
+      "step": 17138
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011156911581569115,
+      "loss": 0.1263,
+      "step": 17139
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0001115566625155666,
+      "loss": 0.1443,
+      "step": 17140
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011154420921544208,
+      "loss": 0.1975,
+      "step": 17141
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011153175591531755,
+      "loss": 0.2526,
+      "step": 17142
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011151930261519301,
+      "loss": 0.141,
+      "step": 17143
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011150684931506848,
+      "loss": 0.123,
+      "step": 17144
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011149439601494395,
+      "loss": 0.1036,
+      "step": 17145
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011148194271481941,
+      "loss": 0.1382,
+      "step": 17146
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011146948941469489,
+      "loss": 0.1197,
+      "step": 17147
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011145703611457035,
+      "loss": 0.3332,
+      "step": 17148
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011144458281444582,
+      "loss": 0.1639,
+      "step": 17149
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011143212951432129,
+      "loss": 0.2698,
+      "step": 17150
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011141967621419675,
+      "loss": 0.0883,
+      "step": 17151
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011140722291407222,
+      "loss": 0.1898,
+      "step": 17152
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0001113947696139477,
+      "loss": 0.1601,
+      "step": 17153
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00011138231631382316,
+      "loss": 0.3081,
+      "step": 17154
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011136986301369861,
+      "loss": 0.2502,
+      "step": 17155
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011135740971357409,
+      "loss": 0.2027,
+      "step": 17156
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011134495641344955,
+      "loss": 0.1773,
+      "step": 17157
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011133250311332502,
+      "loss": 0.2207,
+      "step": 17158
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011132004981320048,
+      "loss": 0.1155,
+      "step": 17159
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011130759651307595,
+      "loss": 0.1208,
+      "step": 17160
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011129514321295142,
+      "loss": 0.1977,
+      "step": 17161
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011128268991282688,
+      "loss": 0.2306,
+      "step": 17162
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011127023661270236,
+      "loss": 0.1088,
+      "step": 17163
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011125778331257783,
+      "loss": 0.1602,
+      "step": 17164
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011124533001245329,
+      "loss": 0.1185,
+      "step": 17165
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011123287671232876,
+      "loss": 0.1641,
+      "step": 17166
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011122042341220423,
+      "loss": 0.2512,
+      "step": 17167
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011120797011207969,
+      "loss": 0.1539,
+      "step": 17168
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011119551681195516,
+      "loss": 0.1607,
+      "step": 17169
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011118306351183064,
+      "loss": 0.1235,
+      "step": 17170
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011117061021170608,
+      "loss": 0.1223,
+      "step": 17171
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011115815691158156,
+      "loss": 0.2609,
+      "step": 17172
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011114570361145702,
+      "loss": 0.2342,
+      "step": 17173
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011113325031133249,
+      "loss": 0.2024,
+      "step": 17174
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011112079701120796,
+      "loss": 0.1881,
+      "step": 17175
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011110834371108342,
+      "loss": 0.1732,
+      "step": 17176
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011109589041095889,
+      "loss": 0.0971,
+      "step": 17177
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011108343711083437,
+      "loss": 0.2347,
+      "step": 17178
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011107098381070982,
+      "loss": 0.2864,
+      "step": 17179
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0001110585305105853,
+      "loss": 0.1958,
+      "step": 17180
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011104607721046077,
+      "loss": 0.2229,
+      "step": 17181
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011103362391033623,
+      "loss": 0.1495,
+      "step": 17182
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001110211706102117,
+      "loss": 0.1071,
+      "step": 17183
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011100871731008717,
+      "loss": 0.2628,
+      "step": 17184
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011099626400996263,
+      "loss": 0.204,
+      "step": 17185
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011098381070983809,
+      "loss": 0.2921,
+      "step": 17186
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011097135740971355,
+      "loss": 0.2911,
+      "step": 17187
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011095890410958902,
+      "loss": 0.2157,
+      "step": 17188
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001109464508094645,
+      "loss": 0.1581,
+      "step": 17189
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011093399750933996,
+      "loss": 0.1404,
+      "step": 17190
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011092154420921543,
+      "loss": 0.2806,
+      "step": 17191
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001109090909090909,
+      "loss": 0.2639,
+      "step": 17192
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011089663760896636,
+      "loss": 0.2016,
+      "step": 17193
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011088418430884183,
+      "loss": 0.1687,
+      "step": 17194
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001108717310087173,
+      "loss": 0.1597,
+      "step": 17195
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011085927770859277,
+      "loss": 0.1711,
+      "step": 17196
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011084682440846824,
+      "loss": 0.1212,
+      "step": 17197
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011083437110834371,
+      "loss": 0.2125,
+      "step": 17198
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011082191780821917,
+      "loss": 0.1676,
+      "step": 17199
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011080946450809464,
+      "loss": 0.2653,
+      "step": 17200
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011079701120797012,
+      "loss": 0.1275,
+      "step": 17201
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011078455790784556,
+      "loss": 0.1075,
+      "step": 17202
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011077210460772103,
+      "loss": 0.1236,
+      "step": 17203
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011075965130759649,
+      "loss": 0.1797,
+      "step": 17204
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011074719800747197,
+      "loss": 0.2654,
+      "step": 17205
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011073474470734744,
+      "loss": 0.1639,
+      "step": 17206
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001107222914072229,
+      "loss": 0.1821,
+      "step": 17207
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011070983810709837,
+      "loss": 0.109,
+      "step": 17208
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011069738480697384,
+      "loss": 0.1582,
+      "step": 17209
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001106849315068493,
+      "loss": 0.2449,
+      "step": 17210
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011067247820672478,
+      "loss": 0.2733,
+      "step": 17211
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011066002490660025,
+      "loss": 0.204,
+      "step": 17212
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011064757160647571,
+      "loss": 0.2031,
+      "step": 17213
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011063511830635118,
+      "loss": 0.1797,
+      "step": 17214
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011062266500622665,
+      "loss": 0.1282,
+      "step": 17215
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011061021170610211,
+      "loss": 0.2049,
+      "step": 17216
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011059775840597758,
+      "loss": 0.2494,
+      "step": 17217
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011058530510585303,
+      "loss": 0.1785,
+      "step": 17218
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0001105728518057285,
+      "loss": 0.193,
+      "step": 17219
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011056039850560398,
+      "loss": 0.2111,
+      "step": 17220
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011054794520547943,
+      "loss": 0.1176,
+      "step": 17221
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011053549190535491,
+      "loss": 0.1804,
+      "step": 17222
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011052303860523038,
+      "loss": 0.2177,
+      "step": 17223
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011051058530510584,
+      "loss": 0.2345,
+      "step": 17224
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011049813200498131,
+      "loss": 0.1735,
+      "step": 17225
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011048567870485678,
+      "loss": 0.1406,
+      "step": 17226
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011047322540473224,
+      "loss": 0.1696,
+      "step": 17227
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011046077210460772,
+      "loss": 0.2862,
+      "step": 17228
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011044831880448318,
+      "loss": 0.3587,
+      "step": 17229
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011043586550435865,
+      "loss": 0.3024,
+      "step": 17230
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011042341220423412,
+      "loss": 0.16,
+      "step": 17231
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011041095890410958,
+      "loss": 0.1313,
+      "step": 17232
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011039850560398505,
+      "loss": 0.1625,
+      "step": 17233
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011038605230386051,
+      "loss": 0.159,
+      "step": 17234
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011037359900373597,
+      "loss": 0.1987,
+      "step": 17235
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011036114570361144,
+      "loss": 0.2577,
+      "step": 17236
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011034869240348692,
+      "loss": 0.1838,
+      "step": 17237
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011033623910336238,
+      "loss": 0.18,
+      "step": 17238
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011032378580323785,
+      "loss": 0.2222,
+      "step": 17239
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011031133250311331,
+      "loss": 0.1502,
+      "step": 17240
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011029887920298878,
+      "loss": 0.1692,
+      "step": 17241
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011028642590286425,
+      "loss": 0.2591,
+      "step": 17242
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011027397260273971,
+      "loss": 0.1857,
+      "step": 17243
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011026151930261519,
+      "loss": 0.1343,
+      "step": 17244
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011024906600249066,
+      "loss": 0.1218,
+      "step": 17245
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011023661270236612,
+      "loss": 0.1277,
+      "step": 17246
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011022415940224159,
+      "loss": 0.1698,
+      "step": 17247
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011021170610211706,
+      "loss": 0.2135,
+      "step": 17248
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011019925280199251,
+      "loss": 0.1491,
+      "step": 17249
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011018679950186798,
+      "loss": 0.1176,
+      "step": 17250
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011017434620174344,
+      "loss": 0.2505,
+      "step": 17251
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011016189290161891,
+      "loss": 0.152,
+      "step": 17252
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011014943960149439,
+      "loss": 0.1425,
+      "step": 17253
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011013698630136984,
+      "loss": 0.2007,
+      "step": 17254
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011012453300124532,
+      "loss": 0.1825,
+      "step": 17255
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011011207970112079,
+      "loss": 0.1143,
+      "step": 17256
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011009962640099625,
+      "loss": 0.1302,
+      "step": 17257
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00011008717310087172,
+      "loss": 0.149,
+      "step": 17258
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0001100747198007472,
+      "loss": 0.1658,
+      "step": 17259
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00011006226650062265,
+      "loss": 0.2519,
+      "step": 17260
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00011004981320049813,
+      "loss": 0.2026,
+      "step": 17261
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0001100373599003736,
+      "loss": 0.1503,
+      "step": 17262
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00011002490660024906,
+      "loss": 0.1514,
+      "step": 17263
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00011001245330012453,
+      "loss": 0.1532,
+      "step": 17264
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010999999999999998,
+      "loss": 0.1174,
+      "step": 17265
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010998754669987545,
+      "loss": 0.1599,
+      "step": 17266
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010997509339975092,
+      "loss": 0.2861,
+      "step": 17267
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010996264009962638,
+      "loss": 0.192,
+      "step": 17268
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010995018679950185,
+      "loss": 0.1549,
+      "step": 17269
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010993773349937733,
+      "loss": 0.1807,
+      "step": 17270
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010992528019925279,
+      "loss": 0.1298,
+      "step": 17271
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010991282689912826,
+      "loss": 0.1304,
+      "step": 17272
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010990037359900373,
+      "loss": 0.2375,
+      "step": 17273
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010988792029887919,
+      "loss": 0.219,
+      "step": 17274
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010987546699875466,
+      "loss": 0.1181,
+      "step": 17275
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010986301369863014,
+      "loss": 0.1213,
+      "step": 17276
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0001098505603985056,
+      "loss": 0.1172,
+      "step": 17277
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010983810709838107,
+      "loss": 0.088,
+      "step": 17278
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010982565379825654,
+      "loss": 0.2157,
+      "step": 17279
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.000109813200498132,
+      "loss": 0.2045,
+      "step": 17280
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010980074719800746,
+      "loss": 0.1581,
+      "step": 17281
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010978829389788292,
+      "loss": 0.2091,
+      "step": 17282
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010977584059775839,
+      "loss": 0.1218,
+      "step": 17283
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00010976338729763386,
+      "loss": 0.1328,
+      "step": 17284
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010975093399750932,
+      "loss": 0.1311,
+      "step": 17285
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0001097384806973848,
+      "loss": 0.1699,
+      "step": 17286
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010972602739726027,
+      "loss": 0.2326,
+      "step": 17287
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010971357409713573,
+      "loss": 0.141,
+      "step": 17288
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0001097011207970112,
+      "loss": 0.1659,
+      "step": 17289
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010968866749688667,
+      "loss": 0.1394,
+      "step": 17290
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010967621419676213,
+      "loss": 0.1367,
+      "step": 17291
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0001096637608966376,
+      "loss": 0.2559,
+      "step": 17292
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010965130759651308,
+      "loss": 0.2831,
+      "step": 17293
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010963885429638854,
+      "loss": 0.1238,
+      "step": 17294
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010962640099626401,
+      "loss": 0.0957,
+      "step": 17295
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010961394769613945,
+      "loss": 0.1432,
+      "step": 17296
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010960149439601493,
+      "loss": 0.1228,
+      "step": 17297
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0001095890410958904,
+      "loss": 0.1982,
+      "step": 17298
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010957658779576586,
+      "loss": 0.2657,
+      "step": 17299
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010956413449564133,
+      "loss": 0.1915,
+      "step": 17300
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0001095516811955168,
+      "loss": 0.1695,
+      "step": 17301
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010953922789539226,
+      "loss": 0.1376,
+      "step": 17302
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010952677459526774,
+      "loss": 0.2357,
+      "step": 17303
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010951432129514321,
+      "loss": 0.2305,
+      "step": 17304
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010950186799501867,
+      "loss": 0.2676,
+      "step": 17305
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010948941469489414,
+      "loss": 0.1707,
+      "step": 17306
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010947696139476961,
+      "loss": 0.2863,
+      "step": 17307
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010946450809464507,
+      "loss": 0.1413,
+      "step": 17308
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00010945205479452055,
+      "loss": 0.1428,
+      "step": 17309
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.000109439601494396,
+      "loss": 0.133,
+      "step": 17310
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010942714819427148,
+      "loss": 0.1936,
+      "step": 17311
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010941469489414694,
+      "loss": 0.1876,
+      "step": 17312
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001094022415940224,
+      "loss": 0.1886,
+      "step": 17313
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010938978829389787,
+      "loss": 0.1303,
+      "step": 17314
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010937733499377334,
+      "loss": 0.1602,
+      "step": 17315
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001093648816936488,
+      "loss": 0.2341,
+      "step": 17316
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010935242839352427,
+      "loss": 0.2551,
+      "step": 17317
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010933997509339975,
+      "loss": 0.1981,
+      "step": 17318
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001093275217932752,
+      "loss": 0.1235,
+      "step": 17319
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010931506849315068,
+      "loss": 0.2075,
+      "step": 17320
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010930261519302614,
+      "loss": 0.1225,
+      "step": 17321
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010929016189290161,
+      "loss": 0.247,
+      "step": 17322
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010927770859277708,
+      "loss": 0.2113,
+      "step": 17323
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010926525529265254,
+      "loss": 0.2309,
+      "step": 17324
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010925280199252801,
+      "loss": 0.1861,
+      "step": 17325
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010924034869240349,
+      "loss": 0.1756,
+      "step": 17326
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010922789539227895,
+      "loss": 0.1817,
+      "step": 17327
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0001092154420921544,
+      "loss": 0.1495,
+      "step": 17328
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010920298879202988,
+      "loss": 0.1444,
+      "step": 17329
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010919053549190534,
+      "loss": 0.1526,
+      "step": 17330
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010917808219178081,
+      "loss": 0.195,
+      "step": 17331
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010916562889165627,
+      "loss": 0.1494,
+      "step": 17332
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010915317559153174,
+      "loss": 0.1812,
+      "step": 17333
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010914072229140722,
+      "loss": 0.1266,
+      "step": 17334
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010912826899128267,
+      "loss": 0.2359,
+      "step": 17335
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00010911581569115815,
+      "loss": 0.2749,
+      "step": 17336
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010910336239103362,
+      "loss": 0.1557,
+      "step": 17337
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010909090909090908,
+      "loss": 0.1367,
+      "step": 17338
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010907845579078455,
+      "loss": 0.1427,
+      "step": 17339
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010906600249066002,
+      "loss": 0.1524,
+      "step": 17340
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010905354919053548,
+      "loss": 0.1514,
+      "step": 17341
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010904109589041096,
+      "loss": 0.2924,
+      "step": 17342
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001090286425902864,
+      "loss": 0.2015,
+      "step": 17343
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010901618929016187,
+      "loss": 0.1284,
+      "step": 17344
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010900373599003735,
+      "loss": 0.168,
+      "step": 17345
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0001089912826899128,
+      "loss": 0.1399,
+      "step": 17346
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010897882938978828,
+      "loss": 0.1951,
+      "step": 17347
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010896637608966375,
+      "loss": 0.4652,
+      "step": 17348
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010895392278953921,
+      "loss": 0.2042,
+      "step": 17349
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010894146948941468,
+      "loss": 0.152,
+      "step": 17350
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010892901618929016,
+      "loss": 0.0968,
+      "step": 17351
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010891656288916562,
+      "loss": 0.1509,
+      "step": 17352
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010890410958904109,
+      "loss": 0.1226,
+      "step": 17353
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010889165628891656,
+      "loss": 0.274,
+      "step": 17354
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010887920298879202,
+      "loss": 0.2288,
+      "step": 17355
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010886674968866749,
+      "loss": 0.1465,
+      "step": 17356
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010885429638854297,
+      "loss": 0.1529,
+      "step": 17357
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010884184308841842,
+      "loss": 0.1949,
+      "step": 17358
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010882938978829388,
+      "loss": 0.2115,
+      "step": 17359
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010881693648816934,
+      "loss": 0.1663,
+      "step": 17360
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010880448318804482,
+      "loss": 0.2061,
+      "step": 17361
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00010879202988792029,
+      "loss": 0.2344,
+      "step": 17362
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010877957658779575,
+      "loss": 0.2421,
+      "step": 17363
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010876712328767122,
+      "loss": 0.185,
+      "step": 17364
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001087546699875467,
+      "loss": 0.1624,
+      "step": 17365
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010874221668742215,
+      "loss": 0.1215,
+      "step": 17366
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010872976338729763,
+      "loss": 0.1507,
+      "step": 17367
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001087173100871731,
+      "loss": 0.2838,
+      "step": 17368
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010870485678704856,
+      "loss": 0.1666,
+      "step": 17369
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010869240348692403,
+      "loss": 0.1007,
+      "step": 17370
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001086799501867995,
+      "loss": 0.1401,
+      "step": 17371
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010866749688667496,
+      "loss": 0.1649,
+      "step": 17372
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010865504358655043,
+      "loss": 0.2622,
+      "step": 17373
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010864259028642591,
+      "loss": 0.2033,
+      "step": 17374
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010863013698630135,
+      "loss": 0.2489,
+      "step": 17375
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010861768368617683,
+      "loss": 0.1199,
+      "step": 17376
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010860523038605228,
+      "loss": 0.262,
+      "step": 17377
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010859277708592776,
+      "loss": 0.1932,
+      "step": 17378
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010858032378580323,
+      "loss": 0.2334,
+      "step": 17379
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010856787048567869,
+      "loss": 0.2311,
+      "step": 17380
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010855541718555416,
+      "loss": 0.1594,
+      "step": 17381
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010854296388542963,
+      "loss": 0.1375,
+      "step": 17382
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001085305105853051,
+      "loss": 0.2252,
+      "step": 17383
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010851805728518057,
+      "loss": 0.1179,
+      "step": 17384
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010850560398505604,
+      "loss": 0.1083,
+      "step": 17385
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0001084931506849315,
+      "loss": 0.2069,
+      "step": 17386
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010848069738480697,
+      "loss": 0.1503,
+      "step": 17387
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00010846824408468244,
+      "loss": 0.143,
+      "step": 17388
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001084557907845579,
+      "loss": 0.125,
+      "step": 17389
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010844333748443338,
+      "loss": 0.1413,
+      "step": 17390
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010843088418430882,
+      "loss": 0.0997,
+      "step": 17391
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001084184308841843,
+      "loss": 0.151,
+      "step": 17392
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010840597758405977,
+      "loss": 0.1295,
+      "step": 17393
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010839352428393523,
+      "loss": 0.1822,
+      "step": 17394
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001083810709838107,
+      "loss": 0.2607,
+      "step": 17395
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010836861768368617,
+      "loss": 0.1887,
+      "step": 17396
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010835616438356163,
+      "loss": 0.1258,
+      "step": 17397
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001083437110834371,
+      "loss": 0.2784,
+      "step": 17398
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010833125778331258,
+      "loss": 0.1836,
+      "step": 17399
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010831880448318804,
+      "loss": 0.2423,
+      "step": 17400
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010830635118306351,
+      "loss": 0.1348,
+      "step": 17401
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010829389788293897,
+      "loss": 0.1814,
+      "step": 17402
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010828144458281444,
+      "loss": 0.1565,
+      "step": 17403
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010826899128268991,
+      "loss": 0.2099,
+      "step": 17404
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010825653798256537,
+      "loss": 0.1657,
+      "step": 17405
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010824408468244083,
+      "loss": 0.1588,
+      "step": 17406
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001082316313823163,
+      "loss": 0.2112,
+      "step": 17407
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010821917808219176,
+      "loss": 0.1451,
+      "step": 17408
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010820672478206724,
+      "loss": 0.2125,
+      "step": 17409
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001081942714819427,
+      "loss": 0.1746,
+      "step": 17410
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010818181818181817,
+      "loss": 0.233,
+      "step": 17411
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010816936488169364,
+      "loss": 0.2546,
+      "step": 17412
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0001081569115815691,
+      "loss": 0.141,
+      "step": 17413
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010814445828144457,
+      "loss": 0.1665,
+      "step": 17414
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00010813200498132004,
+      "loss": 0.334,
+      "step": 17415
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001081195516811955,
+      "loss": 0.1722,
+      "step": 17416
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010810709838107098,
+      "loss": 0.2611,
+      "step": 17417
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010809464508094645,
+      "loss": 0.1844,
+      "step": 17418
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010808219178082191,
+      "loss": 0.2019,
+      "step": 17419
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010806973848069738,
+      "loss": 0.1047,
+      "step": 17420
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010805728518057285,
+      "loss": 0.1559,
+      "step": 17421
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001080448318804483,
+      "loss": 0.1537,
+      "step": 17422
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010803237858032377,
+      "loss": 0.3071,
+      "step": 17423
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010801992528019923,
+      "loss": 0.2737,
+      "step": 17424
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0001080074719800747,
+      "loss": 0.2281,
+      "step": 17425
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010799501867995018,
+      "loss": 0.2641,
+      "step": 17426
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010798256537982564,
+      "loss": 0.1285,
+      "step": 17427
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010797011207970111,
+      "loss": 0.117,
+      "step": 17428
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010795765877957658,
+      "loss": 0.2039,
+      "step": 17429
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010794520547945204,
+      "loss": 0.2422,
+      "step": 17430
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010793275217932751,
+      "loss": 0.1576,
+      "step": 17431
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010792029887920299,
+      "loss": 0.2258,
+      "step": 17432
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010790784557907845,
+      "loss": 0.2119,
+      "step": 17433
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010789539227895392,
+      "loss": 0.1461,
+      "step": 17434
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010788293897882939,
+      "loss": 0.1411,
+      "step": 17435
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010787048567870485,
+      "loss": 0.3457,
+      "step": 17436
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010785803237858032,
+      "loss": 0.1706,
+      "step": 17437
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010784557907845577,
+      "loss": 0.2207,
+      "step": 17438
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010783312577833124,
+      "loss": 0.1339,
+      "step": 17439
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010782067247820671,
+      "loss": 0.1718,
+      "step": 17440
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00010780821917808217,
+      "loss": 0.1698,
+      "step": 17441
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010779576587795765,
+      "loss": 0.2879,
+      "step": 17442
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010778331257783312,
+      "loss": 0.176,
+      "step": 17443
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010777085927770858,
+      "loss": 0.2285,
+      "step": 17444
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010775840597758405,
+      "loss": 0.1346,
+      "step": 17445
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010774595267745952,
+      "loss": 0.1611,
+      "step": 17446
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010773349937733498,
+      "loss": 0.1884,
+      "step": 17447
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010772104607721045,
+      "loss": 0.3773,
+      "step": 17448
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010770859277708593,
+      "loss": 0.1465,
+      "step": 17449
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010769613947696139,
+      "loss": 0.1476,
+      "step": 17450
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010768368617683686,
+      "loss": 0.2112,
+      "step": 17451
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010767123287671233,
+      "loss": 0.1336,
+      "step": 17452
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010765877957658778,
+      "loss": 0.2667,
+      "step": 17453
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010764632627646325,
+      "loss": 0.3296,
+      "step": 17454
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010763387297633871,
+      "loss": 0.1678,
+      "step": 17455
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010762141967621418,
+      "loss": 0.189,
+      "step": 17456
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010760896637608966,
+      "loss": 0.2249,
+      "step": 17457
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010759651307596511,
+      "loss": 0.2532,
+      "step": 17458
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010758405977584059,
+      "loss": 0.1265,
+      "step": 17459
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010757160647571606,
+      "loss": 0.1316,
+      "step": 17460
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010755915317559152,
+      "loss": 0.2249,
+      "step": 17461
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010754669987546699,
+      "loss": 0.2064,
+      "step": 17462
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010753424657534246,
+      "loss": 0.1386,
+      "step": 17463
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010752179327521792,
+      "loss": 0.1479,
+      "step": 17464
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0001075093399750934,
+      "loss": 0.1722,
+      "step": 17465
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010749688667496887,
+      "loss": 0.1974,
+      "step": 17466
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00010748443337484433,
+      "loss": 0.2083,
+      "step": 17467
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001074719800747198,
+      "loss": 0.1579,
+      "step": 17468
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010745952677459525,
+      "loss": 0.2001,
+      "step": 17469
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010744707347447072,
+      "loss": 0.1402,
+      "step": 17470
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010743462017434619,
+      "loss": 0.2019,
+      "step": 17471
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010742216687422165,
+      "loss": 0.1903,
+      "step": 17472
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010740971357409712,
+      "loss": 0.2907,
+      "step": 17473
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001073972602739726,
+      "loss": 0.2062,
+      "step": 17474
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010738480697384806,
+      "loss": 0.197,
+      "step": 17475
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010737235367372353,
+      "loss": 0.118,
+      "step": 17476
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.000107359900373599,
+      "loss": 0.1474,
+      "step": 17477
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010734744707347446,
+      "loss": 0.15,
+      "step": 17478
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010733499377334993,
+      "loss": 0.1593,
+      "step": 17479
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001073225404732254,
+      "loss": 0.2358,
+      "step": 17480
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010731008717310086,
+      "loss": 0.1977,
+      "step": 17481
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010729763387297634,
+      "loss": 0.1331,
+      "step": 17482
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0001072851805728518,
+      "loss": 0.1699,
+      "step": 17483
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010727272727272727,
+      "loss": 0.1511,
+      "step": 17484
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010726027397260273,
+      "loss": 0.1461,
+      "step": 17485
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010724782067247819,
+      "loss": 0.2512,
+      "step": 17486
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010723536737235366,
+      "loss": 0.136,
+      "step": 17487
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010722291407222913,
+      "loss": 0.1671,
+      "step": 17488
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010721046077210459,
+      "loss": 0.098,
+      "step": 17489
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010719800747198007,
+      "loss": 0.1131,
+      "step": 17490
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010718555417185552,
+      "loss": 0.1353,
+      "step": 17491
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.000107173100871731,
+      "loss": 0.3144,
+      "step": 17492
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.00010716064757160647,
+      "loss": 0.1596,
+      "step": 17493
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010714819427148193,
+      "loss": 0.152,
+      "step": 17494
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001071357409713574,
+      "loss": 0.1294,
+      "step": 17495
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010712328767123287,
+      "loss": 0.2884,
+      "step": 17496
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010711083437110833,
+      "loss": 0.0951,
+      "step": 17497
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001070983810709838,
+      "loss": 0.2068,
+      "step": 17498
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010708592777085928,
+      "loss": 0.1767,
+      "step": 17499
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010707347447073474,
+      "loss": 0.2482,
+      "step": 17500
+    },
+    {
+      "epoch": 6.71,
+      "eval_cer": 0.1022962717287395,
+      "eval_loss": 0.3898713290691376,
+      "eval_runtime": 47.5139,
+      "eval_samples_per_second": 36.894,
+      "eval_steps_per_second": 4.63,
+      "eval_wer": 0.1642756935330588,
+      "step": 17500
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001070610211706102,
+      "loss": 0.144,
+      "step": 17501
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010704856787048566,
+      "loss": 0.1613,
+      "step": 17502
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010703611457036113,
+      "loss": 0.1164,
+      "step": 17503
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001070236612702366,
+      "loss": 0.2935,
+      "step": 17504
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010701120797011206,
+      "loss": 0.3302,
+      "step": 17505
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010699875466998753,
+      "loss": 0.1649,
+      "step": 17506
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.000106986301369863,
+      "loss": 0.1391,
+      "step": 17507
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010697384806973847,
+      "loss": 0.177,
+      "step": 17508
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010696139476961394,
+      "loss": 0.1281,
+      "step": 17509
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010694894146948941,
+      "loss": 0.2235,
+      "step": 17510
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010693648816936487,
+      "loss": 0.2,
+      "step": 17511
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010692403486924034,
+      "loss": 0.1961,
+      "step": 17512
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010691158156911582,
+      "loss": 0.1391,
+      "step": 17513
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010689912826899127,
+      "loss": 0.1891,
+      "step": 17514
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010688667496886675,
+      "loss": 0.1376,
+      "step": 17515
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001068742216687422,
+      "loss": 0.1304,
+      "step": 17516
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010686176836861767,
+      "loss": 0.2036,
+      "step": 17517
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00010684931506849314,
+      "loss": 0.1791,
+      "step": 17518
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0001068368617683686,
+      "loss": 0.1512,
+      "step": 17519
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010682440846824407,
+      "loss": 0.1317,
+      "step": 17520
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010681195516811954,
+      "loss": 0.1038,
+      "step": 17521
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.000106799501867995,
+      "loss": 0.1746,
+      "step": 17522
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010678704856787048,
+      "loss": 0.2463,
+      "step": 17523
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010677459526774595,
+      "loss": 0.2151,
+      "step": 17524
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010676214196762141,
+      "loss": 0.1952,
+      "step": 17525
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010674968866749688,
+      "loss": 0.0963,
+      "step": 17526
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010673723536737235,
+      "loss": 0.1339,
+      "step": 17527
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010672478206724781,
+      "loss": 0.1611,
+      "step": 17528
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010671232876712328,
+      "loss": 0.1968,
+      "step": 17529
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010669987546699876,
+      "loss": 0.2638,
+      "step": 17530
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010668742216687422,
+      "loss": 0.1436,
+      "step": 17531
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010667496886674968,
+      "loss": 0.1192,
+      "step": 17532
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010666251556662513,
+      "loss": 0.1367,
+      "step": 17533
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010665006226650061,
+      "loss": 0.1308,
+      "step": 17534
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010663760896637608,
+      "loss": 0.1365,
+      "step": 17535
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010662515566625154,
+      "loss": 0.2903,
+      "step": 17536
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010661270236612701,
+      "loss": 0.2217,
+      "step": 17537
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010660024906600248,
+      "loss": 0.1976,
+      "step": 17538
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010658779576587794,
+      "loss": 0.1473,
+      "step": 17539
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010657534246575342,
+      "loss": 0.1251,
+      "step": 17540
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010656288916562889,
+      "loss": 0.1925,
+      "step": 17541
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010655043586550435,
+      "loss": 0.2243,
+      "step": 17542
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010653798256537982,
+      "loss": 0.1299,
+      "step": 17543
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0001065255292652553,
+      "loss": 0.1213,
+      "step": 17544
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00010651307596513075,
+      "loss": 0.205,
+      "step": 17545
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010650062266500623,
+      "loss": 0.1503,
+      "step": 17546
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001064881693648817,
+      "loss": 0.1232,
+      "step": 17547
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010647571606475714,
+      "loss": 0.2109,
+      "step": 17548
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010646326276463262,
+      "loss": 0.2495,
+      "step": 17549
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010645080946450808,
+      "loss": 0.2741,
+      "step": 17550
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010643835616438355,
+      "loss": 0.2052,
+      "step": 17551
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010642590286425902,
+      "loss": 0.176,
+      "step": 17552
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010641344956413448,
+      "loss": 0.1807,
+      "step": 17553
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010640099626400995,
+      "loss": 0.1404,
+      "step": 17554
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010638854296388543,
+      "loss": 0.1684,
+      "step": 17555
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010637608966376089,
+      "loss": 0.3312,
+      "step": 17556
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010636363636363636,
+      "loss": 0.1409,
+      "step": 17557
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010635118306351183,
+      "loss": 0.199,
+      "step": 17558
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010633872976338729,
+      "loss": 0.2011,
+      "step": 17559
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010632627646326276,
+      "loss": 0.1447,
+      "step": 17560
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010631382316313824,
+      "loss": 0.2525,
+      "step": 17561
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001063013698630137,
+      "loss": 0.15,
+      "step": 17562
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010628891656288915,
+      "loss": 0.1895,
+      "step": 17563
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010627646326276461,
+      "loss": 0.1433,
+      "step": 17564
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010626400996264009,
+      "loss": 0.1567,
+      "step": 17565
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010625155666251556,
+      "loss": 0.1465,
+      "step": 17566
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010623910336239102,
+      "loss": 0.2064,
+      "step": 17567
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010622665006226649,
+      "loss": 0.238,
+      "step": 17568
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010621419676214196,
+      "loss": 0.1341,
+      "step": 17569
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00010620174346201742,
+      "loss": 0.1982,
+      "step": 17570
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0001061892901618929,
+      "loss": 0.1205,
+      "step": 17571
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010617683686176835,
+      "loss": 0.1097,
+      "step": 17572
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010616438356164383,
+      "loss": 0.1685,
+      "step": 17573
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001061519302615193,
+      "loss": 0.1602,
+      "step": 17574
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010613947696139476,
+      "loss": 0.1648,
+      "step": 17575
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010612702366127023,
+      "loss": 0.2295,
+      "step": 17576
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001061145703611457,
+      "loss": 0.2282,
+      "step": 17577
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010610211706102116,
+      "loss": 0.1114,
+      "step": 17578
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010608966376089662,
+      "loss": 0.1351,
+      "step": 17579
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001060772104607721,
+      "loss": 0.157,
+      "step": 17580
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010606475716064755,
+      "loss": 0.1503,
+      "step": 17581
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010605230386052303,
+      "loss": 0.2304,
+      "step": 17582
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010603985056039849,
+      "loss": 0.1794,
+      "step": 17583
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010602739726027396,
+      "loss": 0.1579,
+      "step": 17584
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010601494396014943,
+      "loss": 0.2439,
+      "step": 17585
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010600249066002489,
+      "loss": 0.1923,
+      "step": 17586
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010599003735990036,
+      "loss": 0.1626,
+      "step": 17587
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010597758405977584,
+      "loss": 0.1613,
+      "step": 17588
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001059651307596513,
+      "loss": 0.1251,
+      "step": 17589
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010595267745952677,
+      "loss": 0.154,
+      "step": 17590
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010594022415940224,
+      "loss": 0.0809,
+      "step": 17591
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001059277708592777,
+      "loss": 0.1815,
+      "step": 17592
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010591531755915317,
+      "loss": 0.2115,
+      "step": 17593
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010590286425902865,
+      "loss": 0.1813,
+      "step": 17594
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010589041095890409,
+      "loss": 0.139,
+      "step": 17595
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010587795765877956,
+      "loss": 0.2067,
+      "step": 17596
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00010586550435865502,
+      "loss": 0.1369,
+      "step": 17597
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001058530510585305,
+      "loss": 0.2038,
+      "step": 17598
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010584059775840597,
+      "loss": 0.1557,
+      "step": 17599
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010582814445828143,
+      "loss": 0.2558,
+      "step": 17600
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001058156911581569,
+      "loss": 0.2068,
+      "step": 17601
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010580323785803237,
+      "loss": 0.1236,
+      "step": 17602
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010579078455790783,
+      "loss": 0.0944,
+      "step": 17603
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001057783312577833,
+      "loss": 0.2285,
+      "step": 17604
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010576587795765878,
+      "loss": 0.1889,
+      "step": 17605
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010575342465753424,
+      "loss": 0.2338,
+      "step": 17606
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010574097135740971,
+      "loss": 0.218,
+      "step": 17607
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010572851805728518,
+      "loss": 0.2162,
+      "step": 17608
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010571606475716064,
+      "loss": 0.1693,
+      "step": 17609
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001057036114570361,
+      "loss": 0.1644,
+      "step": 17610
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010569115815691156,
+      "loss": 0.2493,
+      "step": 17611
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010567870485678703,
+      "loss": 0.1532,
+      "step": 17612
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0001056662515566625,
+      "loss": 0.2134,
+      "step": 17613
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010565379825653796,
+      "loss": 0.1443,
+      "step": 17614
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010564134495641344,
+      "loss": 0.1329,
+      "step": 17615
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010562889165628891,
+      "loss": 0.1386,
+      "step": 17616
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010561643835616437,
+      "loss": 0.2408,
+      "step": 17617
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010560398505603984,
+      "loss": 0.1843,
+      "step": 17618
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010559153175591531,
+      "loss": 0.1816,
+      "step": 17619
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010557907845579077,
+      "loss": 0.1647,
+      "step": 17620
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010556662515566625,
+      "loss": 0.0968,
+      "step": 17621
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010555417185554172,
+      "loss": 0.1319,
+      "step": 17622
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.00010554171855541718,
+      "loss": 0.1868,
+      "step": 17623
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010552926525529265,
+      "loss": 0.3285,
+      "step": 17624
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010551681195516812,
+      "loss": 0.2019,
+      "step": 17625
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010550435865504357,
+      "loss": 0.1406,
+      "step": 17626
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010549190535491904,
+      "loss": 0.2575,
+      "step": 17627
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0001054794520547945,
+      "loss": 0.1951,
+      "step": 17628
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010546699875466997,
+      "loss": 0.1448,
+      "step": 17629
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010545454545454545,
+      "loss": 0.248,
+      "step": 17630
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0001054420921544209,
+      "loss": 0.1527,
+      "step": 17631
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010542963885429638,
+      "loss": 0.2137,
+      "step": 17632
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010541718555417185,
+      "loss": 0.2725,
+      "step": 17633
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010540473225404731,
+      "loss": 0.1518,
+      "step": 17634
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010539227895392278,
+      "loss": 0.1087,
+      "step": 17635
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010537982565379826,
+      "loss": 0.3281,
+      "step": 17636
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010536737235367371,
+      "loss": 0.1743,
+      "step": 17637
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010535491905354919,
+      "loss": 0.122,
+      "step": 17638
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010534246575342466,
+      "loss": 0.156,
+      "step": 17639
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010533001245330012,
+      "loss": 0.1624,
+      "step": 17640
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010531755915317559,
+      "loss": 0.137,
+      "step": 17641
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010530510585305104,
+      "loss": 0.339,
+      "step": 17642
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010529265255292651,
+      "loss": 0.3316,
+      "step": 17643
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010528019925280198,
+      "loss": 0.1524,
+      "step": 17644
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010526774595267744,
+      "loss": 0.1566,
+      "step": 17645
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010525529265255292,
+      "loss": 0.1777,
+      "step": 17646
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010524283935242839,
+      "loss": 0.2499,
+      "step": 17647
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010523038605230385,
+      "loss": 0.2048,
+      "step": 17648
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010521793275217932,
+      "loss": 0.2352,
+      "step": 17649
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010520547945205479,
+      "loss": 0.1658,
+      "step": 17650
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010519302615193025,
+      "loss": 0.1311,
+      "step": 17651
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010518057285180572,
+      "loss": 0.2403,
+      "step": 17652
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010516811955168118,
+      "loss": 0.15,
+      "step": 17653
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010515566625155666,
+      "loss": 0.2713,
+      "step": 17654
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010514321295143213,
+      "loss": 0.2044,
+      "step": 17655
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010513075965130759,
+      "loss": 0.1957,
+      "step": 17656
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010511830635118306,
+      "loss": 0.1384,
+      "step": 17657
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010510585305105852,
+      "loss": 0.211,
+      "step": 17658
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010509339975093398,
+      "loss": 0.1312,
+      "step": 17659
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010508094645080945,
+      "loss": 0.2196,
+      "step": 17660
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010506849315068492,
+      "loss": 0.2368,
+      "step": 17661
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010505603985056038,
+      "loss": 0.1375,
+      "step": 17662
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010504358655043586,
+      "loss": 0.1808,
+      "step": 17663
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010503113325031132,
+      "loss": 0.296,
+      "step": 17664
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010501867995018679,
+      "loss": 0.214,
+      "step": 17665
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010500622665006226,
+      "loss": 0.1516,
+      "step": 17666
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010499377334993772,
+      "loss": 0.1814,
+      "step": 17667
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010498132004981319,
+      "loss": 0.1622,
+      "step": 17668
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010496886674968867,
+      "loss": 0.1153,
+      "step": 17669
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010495641344956412,
+      "loss": 0.2062,
+      "step": 17670
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0001049439601494396,
+      "loss": 0.2045,
+      "step": 17671
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010493150684931507,
+      "loss": 0.2243,
+      "step": 17672
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010491905354919052,
+      "loss": 0.2817,
+      "step": 17673
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010490660024906599,
+      "loss": 0.1537,
+      "step": 17674
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010489414694894145,
+      "loss": 0.1489,
+      "step": 17675
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010488169364881692,
+      "loss": 0.1404,
+      "step": 17676
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0001048692403486924,
+      "loss": 0.2915,
+      "step": 17677
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010485678704856785,
+      "loss": 0.1813,
+      "step": 17678
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010484433374844333,
+      "loss": 0.2874,
+      "step": 17679
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0001048318804483188,
+      "loss": 0.2518,
+      "step": 17680
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010481942714819426,
+      "loss": 0.1432,
+      "step": 17681
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010480697384806973,
+      "loss": 0.1546,
+      "step": 17682
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0001047945205479452,
+      "loss": 0.1712,
+      "step": 17683
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010478206724782066,
+      "loss": 0.1875,
+      "step": 17684
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010476961394769613,
+      "loss": 0.2315,
+      "step": 17685
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010475716064757161,
+      "loss": 0.2323,
+      "step": 17686
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010474470734744707,
+      "loss": 0.1744,
+      "step": 17687
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010473225404732254,
+      "loss": 0.1742,
+      "step": 17688
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010471980074719798,
+      "loss": 0.0941,
+      "step": 17689
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010470734744707346,
+      "loss": 0.1921,
+      "step": 17690
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010469489414694893,
+      "loss": 0.2442,
+      "step": 17691
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010468244084682439,
+      "loss": 0.2275,
+      "step": 17692
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010466998754669986,
+      "loss": 0.1295,
+      "step": 17693
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010465753424657533,
+      "loss": 0.1779,
+      "step": 17694
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0001046450809464508,
+      "loss": 0.1417,
+      "step": 17695
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010463262764632627,
+      "loss": 0.254,
+      "step": 17696
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010462017434620174,
+      "loss": 0.1792,
+      "step": 17697
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0001046077210460772,
+      "loss": 0.1767,
+      "step": 17698
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010459526774595267,
+      "loss": 0.148,
+      "step": 17699
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010458281444582814,
+      "loss": 0.2154,
+      "step": 17700
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0001045703611457036,
+      "loss": 0.107,
+      "step": 17701
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010455790784557908,
+      "loss": 0.1092,
+      "step": 17702
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010454545454545455,
+      "loss": 0.161,
+      "step": 17703
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010453300124533001,
+      "loss": 0.2067,
+      "step": 17704
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010452054794520547,
+      "loss": 0.2346,
+      "step": 17705
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010450809464508093,
+      "loss": 0.1852,
+      "step": 17706
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001044956413449564,
+      "loss": 0.2272,
+      "step": 17707
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010448318804483187,
+      "loss": 0.1672,
+      "step": 17708
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010447073474470733,
+      "loss": 0.1962,
+      "step": 17709
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0001044582814445828,
+      "loss": 0.2261,
+      "step": 17710
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010444582814445828,
+      "loss": 0.1863,
+      "step": 17711
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010443337484433374,
+      "loss": 0.3179,
+      "step": 17712
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010442092154420921,
+      "loss": 0.178,
+      "step": 17713
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010440846824408468,
+      "loss": 0.2083,
+      "step": 17714
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010439601494396014,
+      "loss": 0.1342,
+      "step": 17715
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010438356164383561,
+      "loss": 0.127,
+      "step": 17716
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010437110834371109,
+      "loss": 0.2899,
+      "step": 17717
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010435865504358654,
+      "loss": 0.1773,
+      "step": 17718
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010434620174346202,
+      "loss": 0.2253,
+      "step": 17719
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010433374844333746,
+      "loss": 0.1521,
+      "step": 17720
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010432129514321294,
+      "loss": 0.1195,
+      "step": 17721
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010430884184308841,
+      "loss": 0.1003,
+      "step": 17722
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010429638854296387,
+      "loss": 0.2524,
+      "step": 17723
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010428393524283934,
+      "loss": 0.2214,
+      "step": 17724
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010427148194271481,
+      "loss": 0.1434,
+      "step": 17725
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010425902864259027,
+      "loss": 0.1273,
+      "step": 17726
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010424657534246574,
+      "loss": 0.1775,
+      "step": 17727
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010423412204234122,
+      "loss": 0.0765,
+      "step": 17728
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010422166874221668,
+      "loss": 0.2383,
+      "step": 17729
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010420921544209215,
+      "loss": 0.2212,
+      "step": 17730
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010419676214196762,
+      "loss": 0.1284,
+      "step": 17731
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010418430884184308,
+      "loss": 0.1493,
+      "step": 17732
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010417185554171855,
+      "loss": 0.1207,
+      "step": 17733
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010415940224159401,
+      "loss": 0.1016,
+      "step": 17734
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010414694894146949,
+      "loss": 0.1601,
+      "step": 17735
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010413449564134495,
+      "loss": 0.2007,
+      "step": 17736
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0001041220423412204,
+      "loss": 0.214,
+      "step": 17737
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010410958904109588,
+      "loss": 0.1496,
+      "step": 17738
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010409713574097135,
+      "loss": 0.1774,
+      "step": 17739
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010408468244084681,
+      "loss": 0.1358,
+      "step": 17740
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010407222914072228,
+      "loss": 0.1731,
+      "step": 17741
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010405977584059775,
+      "loss": 0.1684,
+      "step": 17742
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010404732254047321,
+      "loss": 0.1286,
+      "step": 17743
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010403486924034869,
+      "loss": 0.1284,
+      "step": 17744
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010402241594022415,
+      "loss": 0.1916,
+      "step": 17745
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010400996264009962,
+      "loss": 0.2216,
+      "step": 17746
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010399750933997509,
+      "loss": 0.146,
+      "step": 17747
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010398505603985055,
+      "loss": 0.238,
+      "step": 17748
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010397260273972602,
+      "loss": 0.2067,
+      "step": 17749
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0001039601494396015,
+      "loss": 0.1332,
+      "step": 17750
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010394769613947695,
+      "loss": 0.1149,
+      "step": 17751
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010393524283935241,
+      "loss": 0.1942,
+      "step": 17752
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010392278953922789,
+      "loss": 0.3713,
+      "step": 17753
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.00010391033623910335,
+      "loss": 0.174,
+      "step": 17754
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010389788293897882,
+      "loss": 0.2095,
+      "step": 17755
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010388542963885428,
+      "loss": 0.1812,
+      "step": 17756
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010387297633872975,
+      "loss": 0.2609,
+      "step": 17757
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010386052303860522,
+      "loss": 0.1167,
+      "step": 17758
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010384806973848068,
+      "loss": 0.162,
+      "step": 17759
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010383561643835615,
+      "loss": 0.2217,
+      "step": 17760
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010382316313823163,
+      "loss": 0.2213,
+      "step": 17761
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010381070983810709,
+      "loss": 0.1534,
+      "step": 17762
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010379825653798256,
+      "loss": 0.2005,
+      "step": 17763
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010378580323785803,
+      "loss": 0.1039,
+      "step": 17764
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010377334993773349,
+      "loss": 0.1516,
+      "step": 17765
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010376089663760896,
+      "loss": 0.161,
+      "step": 17766
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010374844333748444,
+      "loss": 0.3191,
+      "step": 17767
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010373599003735988,
+      "loss": 0.2172,
+      "step": 17768
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010372353673723536,
+      "loss": 0.2233,
+      "step": 17769
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010371108343711081,
+      "loss": 0.1386,
+      "step": 17770
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010369863013698629,
+      "loss": 0.1622,
+      "step": 17771
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010368617683686176,
+      "loss": 0.1695,
+      "step": 17772
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010367372353673722,
+      "loss": 0.2567,
+      "step": 17773
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010366127023661269,
+      "loss": 0.2911,
+      "step": 17774
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010364881693648816,
+      "loss": 0.1326,
+      "step": 17775
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010363636363636362,
+      "loss": 0.1774,
+      "step": 17776
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0001036239103362391,
+      "loss": 0.1483,
+      "step": 17777
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010361145703611457,
+      "loss": 0.1735,
+      "step": 17778
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010359900373599003,
+      "loss": 0.2523,
+      "step": 17779
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0001035865504358655,
+      "loss": 0.1816,
+      "step": 17780
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010357409713574097,
+      "loss": 0.2352,
+      "step": 17781
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010356164383561643,
+      "loss": 0.1259,
+      "step": 17782
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010354919053549189,
+      "loss": 0.1253,
+      "step": 17783
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010353673723536735,
+      "loss": 0.1596,
+      "step": 17784
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010352428393524282,
+      "loss": 0.1672,
+      "step": 17785
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0001035118306351183,
+      "loss": 0.28,
+      "step": 17786
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010349937733499376,
+      "loss": 0.2107,
+      "step": 17787
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010348692403486923,
+      "loss": 0.1207,
+      "step": 17788
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0001034744707347447,
+      "loss": 0.138,
+      "step": 17789
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010346201743462016,
+      "loss": 0.1295,
+      "step": 17790
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010344956413449563,
+      "loss": 0.0854,
+      "step": 17791
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0001034371108343711,
+      "loss": 0.1977,
+      "step": 17792
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010342465753424656,
+      "loss": 0.1557,
+      "step": 17793
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010341220423412204,
+      "loss": 0.1311,
+      "step": 17794
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010339975093399751,
+      "loss": 0.1325,
+      "step": 17795
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010338729763387297,
+      "loss": 0.1626,
+      "step": 17796
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010337484433374844,
+      "loss": 0.1406,
+      "step": 17797
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010336239103362392,
+      "loss": 0.1574,
+      "step": 17798
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010334993773349936,
+      "loss": 0.1491,
+      "step": 17799
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010333748443337483,
+      "loss": 0.2094,
+      "step": 17800
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010332503113325029,
+      "loss": 0.1398,
+      "step": 17801
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010331257783312577,
+      "loss": 0.0914,
+      "step": 17802
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010330012453300124,
+      "loss": 0.1607,
+      "step": 17803
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0001032876712328767,
+      "loss": 0.1971,
+      "step": 17804
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010327521793275217,
+      "loss": 0.1634,
+      "step": 17805
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010326276463262764,
+      "loss": 0.1713,
+      "step": 17806
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0001032503113325031,
+      "loss": 0.1538,
+      "step": 17807
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010323785803237857,
+      "loss": 0.1183,
+      "step": 17808
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010322540473225405,
+      "loss": 0.1544,
+      "step": 17809
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0001032129514321295,
+      "loss": 0.1371,
+      "step": 17810
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010320049813200498,
+      "loss": 0.223,
+      "step": 17811
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010318804483188045,
+      "loss": 0.1257,
+      "step": 17812
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010317559153175591,
+      "loss": 0.199,
+      "step": 17813
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010316313823163138,
+      "loss": 0.1062,
+      "step": 17814
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010315068493150683,
+      "loss": 0.1664,
+      "step": 17815
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0001031382316313823,
+      "loss": 0.1438,
+      "step": 17816
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010312577833125777,
+      "loss": 0.2702,
+      "step": 17817
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010311332503113323,
+      "loss": 0.1785,
+      "step": 17818
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0001031008717310087,
+      "loss": 0.2721,
+      "step": 17819
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010308841843088418,
+      "loss": 0.1619,
+      "step": 17820
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010307596513075964,
+      "loss": 0.2433,
+      "step": 17821
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010306351183063511,
+      "loss": 0.1414,
+      "step": 17822
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010305105853051058,
+      "loss": 0.3058,
+      "step": 17823
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010303860523038604,
+      "loss": 0.1672,
+      "step": 17824
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010302615193026152,
+      "loss": 0.1865,
+      "step": 17825
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010301369863013698,
+      "loss": 0.2076,
+      "step": 17826
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010300124533001245,
+      "loss": 0.2197,
+      "step": 17827
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010298879202988792,
+      "loss": 0.218,
+      "step": 17828
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010297633872976338,
+      "loss": 0.1961,
+      "step": 17829
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010296388542963884,
+      "loss": 0.246,
+      "step": 17830
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010295143212951431,
+      "loss": 0.2287,
+      "step": 17831
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010293897882938977,
+      "loss": 0.1654,
+      "step": 17832
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010292652552926524,
+      "loss": 0.1741,
+      "step": 17833
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.0001029140722291407,
+      "loss": 0.1287,
+      "step": 17834
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010290161892901618,
+      "loss": 0.2575,
+      "step": 17835
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010288916562889165,
+      "loss": 0.3054,
+      "step": 17836
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010287671232876711,
+      "loss": 0.1056,
+      "step": 17837
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010286425902864258,
+      "loss": 0.2082,
+      "step": 17838
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010285180572851805,
+      "loss": 0.1823,
+      "step": 17839
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010283935242839351,
+      "loss": 0.1191,
+      "step": 17840
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010282689912826898,
+      "loss": 0.0784,
+      "step": 17841
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010281444582814446,
+      "loss": 0.2172,
+      "step": 17842
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010280199252801992,
+      "loss": 0.1603,
+      "step": 17843
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010278953922789539,
+      "loss": 0.2145,
+      "step": 17844
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010277708592777086,
+      "loss": 0.253,
+      "step": 17845
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010276463262764631,
+      "loss": 0.141,
+      "step": 17846
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010275217932752178,
+      "loss": 0.1566,
+      "step": 17847
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010273972602739724,
+      "loss": 0.3202,
+      "step": 17848
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010272727272727271,
+      "loss": 0.1869,
+      "step": 17849
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010271481942714818,
+      "loss": 0.1473,
+      "step": 17850
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010270236612702364,
+      "loss": 0.1486,
+      "step": 17851
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010268991282689912,
+      "loss": 0.1014,
+      "step": 17852
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010267745952677459,
+      "loss": 0.176,
+      "step": 17853
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010266500622665005,
+      "loss": 0.2137,
+      "step": 17854
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010265255292652552,
+      "loss": 0.2386,
+      "step": 17855
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.000102640099626401,
+      "loss": 0.1649,
+      "step": 17856
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010262764632627645,
+      "loss": 0.17,
+      "step": 17857
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00010261519302615193,
+      "loss": 0.2205,
+      "step": 17858
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0001026027397260274,
+      "loss": 0.0914,
+      "step": 17859
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010259028642590286,
+      "loss": 0.2347,
+      "step": 17860
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010257783312577833,
+      "loss": 0.1732,
+      "step": 17861
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010256537982565378,
+      "loss": 0.142,
+      "step": 17862
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010255292652552925,
+      "loss": 0.1332,
+      "step": 17863
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010254047322540472,
+      "loss": 0.1157,
+      "step": 17864
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010252801992528018,
+      "loss": 0.1011,
+      "step": 17865
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010251556662515565,
+      "loss": 0.1085,
+      "step": 17866
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010250311332503113,
+      "loss": 0.2284,
+      "step": 17867
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010249066002490659,
+      "loss": 0.2015,
+      "step": 17868
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010247820672478206,
+      "loss": 0.2145,
+      "step": 17869
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010246575342465753,
+      "loss": 0.1938,
+      "step": 17870
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010245330012453299,
+      "loss": 0.349,
+      "step": 17871
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010244084682440846,
+      "loss": 0.1294,
+      "step": 17872
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010242839352428394,
+      "loss": 0.344,
+      "step": 17873
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0001024159402241594,
+      "loss": 0.1502,
+      "step": 17874
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010240348692403487,
+      "loss": 0.1447,
+      "step": 17875
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010239103362391034,
+      "loss": 0.2151,
+      "step": 17876
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010237858032378579,
+      "loss": 0.1934,
+      "step": 17877
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010236612702366126,
+      "loss": 0.1379,
+      "step": 17878
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010235367372353672,
+      "loss": 0.2676,
+      "step": 17879
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010234122042341219,
+      "loss": 0.2246,
+      "step": 17880
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010232876712328766,
+      "loss": 0.1537,
+      "step": 17881
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010231631382316312,
+      "loss": 0.1672,
+      "step": 17882
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0001023038605230386,
+      "loss": 0.1935,
+      "step": 17883
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010229140722291407,
+      "loss": 0.1141,
+      "step": 17884
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010227895392278953,
+      "loss": 0.1519,
+      "step": 17885
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.000102266500622665,
+      "loss": 0.257,
+      "step": 17886
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010225404732254047,
+      "loss": 0.2007,
+      "step": 17887
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010224159402241593,
+      "loss": 0.1387,
+      "step": 17888
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0001022291407222914,
+      "loss": 0.1471,
+      "step": 17889
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010221668742216688,
+      "loss": 0.182,
+      "step": 17890
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010220423412204234,
+      "loss": 0.1036,
+      "step": 17891
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010219178082191781,
+      "loss": 0.2491,
+      "step": 17892
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010217932752179325,
+      "loss": 0.2163,
+      "step": 17893
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010216687422166873,
+      "loss": 0.1493,
+      "step": 17894
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0001021544209215442,
+      "loss": 0.0958,
+      "step": 17895
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010214196762141966,
+      "loss": 0.1601,
+      "step": 17896
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010212951432129513,
+      "loss": 0.1048,
+      "step": 17897
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0001021170610211706,
+      "loss": 0.1927,
+      "step": 17898
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010210460772104606,
+      "loss": 0.2211,
+      "step": 17899
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010209215442092154,
+      "loss": 0.1213,
+      "step": 17900
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010207970112079701,
+      "loss": 0.1427,
+      "step": 17901
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010206724782067247,
+      "loss": 0.1181,
+      "step": 17902
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010205479452054794,
+      "loss": 0.135,
+      "step": 17903
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010204234122042341,
+      "loss": 0.252,
+      "step": 17904
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010202988792029887,
+      "loss": 0.1811,
+      "step": 17905
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010201743462017435,
+      "loss": 0.1894,
+      "step": 17906
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0001020049813200498,
+      "loss": 0.1365,
+      "step": 17907
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010199252801992528,
+      "loss": 0.147,
+      "step": 17908
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010198007471980074,
+      "loss": 0.1442,
+      "step": 17909
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0001019676214196762,
+      "loss": 0.1523,
+      "step": 17910
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010195516811955167,
+      "loss": 0.2502,
+      "step": 17911
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010194271481942714,
+      "loss": 0.1918,
+      "step": 17912
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0001019302615193026,
+      "loss": 0.1706,
+      "step": 17913
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010191780821917807,
+      "loss": 0.1074,
+      "step": 17914
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010190535491905353,
+      "loss": 0.183,
+      "step": 17915
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.000101892901618929,
+      "loss": 0.1453,
+      "step": 17916
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010188044831880448,
+      "loss": 0.4869,
+      "step": 17917
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010186799501867994,
+      "loss": 0.1854,
+      "step": 17918
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010185554171855541,
+      "loss": 0.1256,
+      "step": 17919
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010184308841843088,
+      "loss": 0.1919,
+      "step": 17920
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010183063511830634,
+      "loss": 0.1964,
+      "step": 17921
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010181818181818181,
+      "loss": 0.1763,
+      "step": 17922
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010180572851805729,
+      "loss": 0.2608,
+      "step": 17923
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010179327521793275,
+      "loss": 0.186,
+      "step": 17924
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0001017808219178082,
+      "loss": 0.1439,
+      "step": 17925
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010176836861768366,
+      "loss": 0.1179,
+      "step": 17926
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010175591531755914,
+      "loss": 0.081,
+      "step": 17927
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010174346201743461,
+      "loss": 0.2115,
+      "step": 17928
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010173100871731007,
+      "loss": 0.2706,
+      "step": 17929
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010171855541718554,
+      "loss": 0.1562,
+      "step": 17930
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010170610211706101,
+      "loss": 0.1536,
+      "step": 17931
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010169364881693647,
+      "loss": 0.3051,
+      "step": 17932
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010168119551681195,
+      "loss": 0.1948,
+      "step": 17933
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010166874221668742,
+      "loss": 0.1413,
+      "step": 17934
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010165628891656288,
+      "loss": 0.2542,
+      "step": 17935
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010164383561643835,
+      "loss": 0.1605,
+      "step": 17936
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010163138231631382,
+      "loss": 0.281,
+      "step": 17937
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010161892901618928,
+      "loss": 0.2196,
+      "step": 17938
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010160647571606476,
+      "loss": 0.1477,
+      "step": 17939
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0001015940224159402,
+      "loss": 0.2295,
+      "step": 17940
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010158156911581567,
+      "loss": 0.1311,
+      "step": 17941
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010156911581569115,
+      "loss": 0.3198,
+      "step": 17942
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0001015566625155666,
+      "loss": 0.1647,
+      "step": 17943
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010154420921544208,
+      "loss": 0.1762,
+      "step": 17944
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010153175591531755,
+      "loss": 0.1306,
+      "step": 17945
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010151930261519301,
+      "loss": 0.1187,
+      "step": 17946
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010150684931506848,
+      "loss": 0.2069,
+      "step": 17947
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010149439601494396,
+      "loss": 0.2354,
+      "step": 17948
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010148194271481942,
+      "loss": 0.1271,
+      "step": 17949
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010146948941469489,
+      "loss": 0.1498,
+      "step": 17950
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010145703611457036,
+      "loss": 0.1722,
+      "step": 17951
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010144458281444582,
+      "loss": 0.1296,
+      "step": 17952
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010143212951432129,
+      "loss": 0.1873,
+      "step": 17953
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010141967621419677,
+      "loss": 0.259,
+      "step": 17954
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010140722291407222,
+      "loss": 0.2212,
+      "step": 17955
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010139476961394768,
+      "loss": 0.1264,
+      "step": 17956
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010138231631382314,
+      "loss": 0.1334,
+      "step": 17957
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010136986301369862,
+      "loss": 0.1403,
+      "step": 17958
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010135740971357409,
+      "loss": 0.1242,
+      "step": 17959
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010134495641344955,
+      "loss": 0.1671,
+      "step": 17960
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010133250311332502,
+      "loss": 0.3005,
+      "step": 17961
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00010132004981320049,
+      "loss": 0.2731,
+      "step": 17962
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010130759651307595,
+      "loss": 0.1443,
+      "step": 17963
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010129514321295142,
+      "loss": 0.1952,
+      "step": 17964
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001012826899128269,
+      "loss": 0.2079,
+      "step": 17965
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010127023661270236,
+      "loss": 0.1056,
+      "step": 17966
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010125778331257783,
+      "loss": 0.2125,
+      "step": 17967
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001012453300124533,
+      "loss": 0.2331,
+      "step": 17968
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010123287671232876,
+      "loss": 0.1822,
+      "step": 17969
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010122042341220423,
+      "loss": 0.1696,
+      "step": 17970
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001012079701120797,
+      "loss": 0.1094,
+      "step": 17971
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010119551681195515,
+      "loss": 0.1149,
+      "step": 17972
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010118306351183062,
+      "loss": 0.3305,
+      "step": 17973
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010117061021170608,
+      "loss": 0.155,
+      "step": 17974
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010115815691158156,
+      "loss": 0.1505,
+      "step": 17975
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010114570361145703,
+      "loss": 0.145,
+      "step": 17976
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010113325031133249,
+      "loss": 0.1277,
+      "step": 17977
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010112079701120796,
+      "loss": 0.1904,
+      "step": 17978
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010110834371108343,
+      "loss": 0.2986,
+      "step": 17979
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010109589041095889,
+      "loss": 0.2299,
+      "step": 17980
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010108343711083437,
+      "loss": 0.1553,
+      "step": 17981
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010107098381070984,
+      "loss": 0.1121,
+      "step": 17982
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001010585305105853,
+      "loss": 0.1639,
+      "step": 17983
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010104607721046077,
+      "loss": 0.2099,
+      "step": 17984
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010103362391033624,
+      "loss": 0.1683,
+      "step": 17985
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001010211706102117,
+      "loss": 0.2119,
+      "step": 17986
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010100871731008716,
+      "loss": 0.2367,
+      "step": 17987
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00010099626400996262,
+      "loss": 0.1956,
+      "step": 17988
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0001009838107098381,
+      "loss": 0.1366,
+      "step": 17989
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010097135740971357,
+      "loss": 0.1152,
+      "step": 17990
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010095890410958903,
+      "loss": 0.2229,
+      "step": 17991
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001009464508094645,
+      "loss": 0.2303,
+      "step": 17992
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010093399750933997,
+      "loss": 0.1925,
+      "step": 17993
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010092154420921543,
+      "loss": 0.1288,
+      "step": 17994
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001009090909090909,
+      "loss": 0.2002,
+      "step": 17995
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010089663760896636,
+      "loss": 0.1382,
+      "step": 17996
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010088418430884183,
+      "loss": 0.1438,
+      "step": 17997
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010087173100871731,
+      "loss": 0.3238,
+      "step": 17998
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010085927770859277,
+      "loss": 0.2225,
+      "step": 17999
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010084682440846824,
+      "loss": 0.159,
+      "step": 18000
+    },
+    {
+      "epoch": 6.9,
+      "eval_cer": 0.10323663109428956,
+      "eval_loss": 0.3846573531627655,
+      "eval_runtime": 47.6821,
+      "eval_samples_per_second": 36.764,
+      "eval_steps_per_second": 4.614,
+      "eval_wer": 0.16871068470833145,
+      "step": 18000
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010083437110834371,
+      "loss": 0.1246,
+      "step": 18001
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010082191780821917,
+      "loss": 0.2539,
+      "step": 18002
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010080946450809463,
+      "loss": 0.1492,
+      "step": 18003
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001007970112079701,
+      "loss": 0.2471,
+      "step": 18004
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010078455790784556,
+      "loss": 0.1679,
+      "step": 18005
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010077210460772103,
+      "loss": 0.1835,
+      "step": 18006
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001007596513075965,
+      "loss": 0.1472,
+      "step": 18007
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010074719800747197,
+      "loss": 0.1287,
+      "step": 18008
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010073474470734744,
+      "loss": 0.1535,
+      "step": 18009
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001007222914072229,
+      "loss": 0.2,
+      "step": 18010
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010070983810709837,
+      "loss": 0.2046,
+      "step": 18011
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010069738480697384,
+      "loss": 0.2142,
+      "step": 18012
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0001006849315068493,
+      "loss": 0.1315,
+      "step": 18013
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010067247820672478,
+      "loss": 0.099,
+      "step": 18014
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010066002490660025,
+      "loss": 0.1368,
+      "step": 18015
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010064757160647571,
+      "loss": 0.1405,
+      "step": 18016
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010063511830635118,
+      "loss": 0.2118,
+      "step": 18017
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010062266500622665,
+      "loss": 0.1413,
+      "step": 18018
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0001006102117061021,
+      "loss": 0.2742,
+      "step": 18019
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010059775840597757,
+      "loss": 0.1494,
+      "step": 18020
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010058530510585303,
+      "loss": 0.099,
+      "step": 18021
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0001005728518057285,
+      "loss": 0.2806,
+      "step": 18022
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010056039850560398,
+      "loss": 0.1369,
+      "step": 18023
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010054794520547944,
+      "loss": 0.2212,
+      "step": 18024
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010053549190535491,
+      "loss": 0.1345,
+      "step": 18025
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010052303860523038,
+      "loss": 0.1391,
+      "step": 18026
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010051058530510584,
+      "loss": 0.2575,
+      "step": 18027
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010049813200498131,
+      "loss": 0.2168,
+      "step": 18028
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010048567870485679,
+      "loss": 0.0929,
+      "step": 18029
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010047322540473224,
+      "loss": 0.2197,
+      "step": 18030
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010046077210460772,
+      "loss": 0.131,
+      "step": 18031
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010044831880448319,
+      "loss": 0.1413,
+      "step": 18032
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010043586550435865,
+      "loss": 0.1479,
+      "step": 18033
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010042341220423411,
+      "loss": 0.1252,
+      "step": 18034
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010041095890410957,
+      "loss": 0.1752,
+      "step": 18035
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010039850560398504,
+      "loss": 0.2583,
+      "step": 18036
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010038605230386051,
+      "loss": 0.1712,
+      "step": 18037
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010037359900373597,
+      "loss": 0.1547,
+      "step": 18038
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010036114570361144,
+      "loss": 0.1162,
+      "step": 18039
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010034869240348692,
+      "loss": 0.1247,
+      "step": 18040
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010033623910336238,
+      "loss": 0.1113,
+      "step": 18041
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010032378580323785,
+      "loss": 0.343,
+      "step": 18042
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010031133250311332,
+      "loss": 0.2001,
+      "step": 18043
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010029887920298878,
+      "loss": 0.1211,
+      "step": 18044
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010028642590286425,
+      "loss": 0.1915,
+      "step": 18045
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010027397260273973,
+      "loss": 0.1738,
+      "step": 18046
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010026151930261519,
+      "loss": 0.1443,
+      "step": 18047
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010024906600249066,
+      "loss": 0.2541,
+      "step": 18048
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010023661270236613,
+      "loss": 0.3087,
+      "step": 18049
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010022415940224158,
+      "loss": 0.1825,
+      "step": 18050
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010021170610211705,
+      "loss": 0.1697,
+      "step": 18051
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010019925280199251,
+      "loss": 0.2539,
+      "step": 18052
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010018679950186798,
+      "loss": 0.1367,
+      "step": 18053
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010017434620174345,
+      "loss": 0.3581,
+      "step": 18054
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010016189290161891,
+      "loss": 0.259,
+      "step": 18055
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010014943960149439,
+      "loss": 0.2454,
+      "step": 18056
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010013698630136986,
+      "loss": 0.236,
+      "step": 18057
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010012453300124532,
+      "loss": 0.1196,
+      "step": 18058
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010011207970112079,
+      "loss": 0.1679,
+      "step": 18059
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010009962640099626,
+      "loss": 0.1474,
+      "step": 18060
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010008717310087172,
+      "loss": 0.2169,
+      "step": 18061
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0001000747198007472,
+      "loss": 0.1701,
+      "step": 18062
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010006226650062267,
+      "loss": 0.191,
+      "step": 18063
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010004981320049813,
+      "loss": 0.26,
+      "step": 18064
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0001000373599003736,
+      "loss": 0.1644,
+      "step": 18065
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010002490660024905,
+      "loss": 0.1168,
+      "step": 18066
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00010001245330012452,
+      "loss": 0.2558,
+      "step": 18067
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 0.2437,
+      "step": 18068
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.998754669987545e-05,
+      "loss": 0.2318,
+      "step": 18069
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.997509339975092e-05,
+      "loss": 0.2262,
+      "step": 18070
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.99626400996264e-05,
+      "loss": 0.1676,
+      "step": 18071
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.995018679950186e-05,
+      "loss": 0.165,
+      "step": 18072
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.993773349937733e-05,
+      "loss": 0.2145,
+      "step": 18073
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.99252801992528e-05,
+      "loss": 0.2278,
+      "step": 18074
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.991282689912826e-05,
+      "loss": 0.1497,
+      "step": 18075
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.990037359900373e-05,
+      "loss": 0.2006,
+      "step": 18076
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.988792029887919e-05,
+      "loss": 0.116,
+      "step": 18077
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.987546699875466e-05,
+      "loss": 0.1156,
+      "step": 18078
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.986301369863014e-05,
+      "loss": 0.3632,
+      "step": 18079
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.98505603985056e-05,
+      "loss": 0.2234,
+      "step": 18080
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.983810709838107e-05,
+      "loss": 0.2332,
+      "step": 18081
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.982565379825653e-05,
+      "loss": 0.0829,
+      "step": 18082
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.981320049813199e-05,
+      "loss": 0.1209,
+      "step": 18083
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.980074719800746e-05,
+      "loss": 0.1028,
+      "step": 18084
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.978829389788293e-05,
+      "loss": 0.1485,
+      "step": 18085
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.977584059775839e-05,
+      "loss": 0.3512,
+      "step": 18086
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.976338729763386e-05,
+      "loss": 0.1347,
+      "step": 18087
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.975093399750932e-05,
+      "loss": 0.1606,
+      "step": 18088
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.97384806973848e-05,
+      "loss": 0.1532,
+      "step": 18089
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.972602739726027e-05,
+      "loss": 0.1188,
+      "step": 18090
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.971357409713573e-05,
+      "loss": 0.1097,
+      "step": 18091
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.97011207970112e-05,
+      "loss": 0.2088,
+      "step": 18092
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.968866749688667e-05,
+      "loss": 0.2832,
+      "step": 18093
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.967621419676213e-05,
+      "loss": 0.1246,
+      "step": 18094
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.96637608966376e-05,
+      "loss": 0.219,
+      "step": 18095
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.965130759651308e-05,
+      "loss": 0.1538,
+      "step": 18096
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.963885429638852e-05,
+      "loss": 0.1703,
+      "step": 18097
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.9626400996264e-05,
+      "loss": 0.1225,
+      "step": 18098
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.961394769613946e-05,
+      "loss": 0.2029,
+      "step": 18099
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.960149439601493e-05,
+      "loss": 0.151,
+      "step": 18100
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.95890410958904e-05,
+      "loss": 0.2017,
+      "step": 18101
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.957658779576586e-05,
+      "loss": 0.1404,
+      "step": 18102
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.956413449564133e-05,
+      "loss": 0.0989,
+      "step": 18103
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.95516811955168e-05,
+      "loss": 0.1458,
+      "step": 18104
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.953922789539227e-05,
+      "loss": 0.2278,
+      "step": 18105
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.952677459526774e-05,
+      "loss": 0.1206,
+      "step": 18106
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.951432129514321e-05,
+      "loss": 0.1404,
+      "step": 18107
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.950186799501867e-05,
+      "loss": 0.1208,
+      "step": 18108
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.948941469489414e-05,
+      "loss": 0.1584,
+      "step": 18109
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.947696139476962e-05,
+      "loss": 0.179,
+      "step": 18110
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.946450809464507e-05,
+      "loss": 0.2799,
+      "step": 18111
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.945205479452055e-05,
+      "loss": 0.2016,
+      "step": 18112
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.943960149439599e-05,
+      "loss": 0.1289,
+      "step": 18113
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.942714819427147e-05,
+      "loss": 0.1606,
+      "step": 18114
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.941469489414694e-05,
+      "loss": 0.1698,
+      "step": 18115
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.94022415940224e-05,
+      "loss": 0.0943,
+      "step": 18116
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.938978829389787e-05,
+      "loss": 0.253,
+      "step": 18117
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.937733499377334e-05,
+      "loss": 0.1379,
+      "step": 18118
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.93648816936488e-05,
+      "loss": 0.1509,
+      "step": 18119
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.935242839352427e-05,
+      "loss": 0.1224,
+      "step": 18120
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.933997509339975e-05,
+      "loss": 0.1085,
+      "step": 18121
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.93275217932752e-05,
+      "loss": 0.2106,
+      "step": 18122
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.931506849315068e-05,
+      "loss": 0.2083,
+      "step": 18123
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.930261519302615e-05,
+      "loss": 0.2629,
+      "step": 18124
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.929016189290161e-05,
+      "loss": 0.2176,
+      "step": 18125
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.927770859277708e-05,
+      "loss": 0.1366,
+      "step": 18126
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.926525529265256e-05,
+      "loss": 0.1055,
+      "step": 18127
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.925280199252802e-05,
+      "loss": 0.155,
+      "step": 18128
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.924034869240347e-05,
+      "loss": 0.3214,
+      "step": 18129
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.922789539227893e-05,
+      "loss": 0.1534,
+      "step": 18130
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.92154420921544e-05,
+      "loss": 0.2174,
+      "step": 18131
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.920298879202988e-05,
+      "loss": 0.1539,
+      "step": 18132
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.919053549190534e-05,
+      "loss": 0.1484,
+      "step": 18133
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.917808219178081e-05,
+      "loss": 0.1419,
+      "step": 18134
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.916562889165628e-05,
+      "loss": 0.175,
+      "step": 18135
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.915317559153174e-05,
+      "loss": 0.256,
+      "step": 18136
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.914072229140722e-05,
+      "loss": 0.2472,
+      "step": 18137
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.912826899128269e-05,
+      "loss": 0.2478,
+      "step": 18138
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.911581569115815e-05,
+      "loss": 0.2199,
+      "step": 18139
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.910336239103362e-05,
+      "loss": 0.1785,
+      "step": 18140
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.90909090909091e-05,
+      "loss": 0.2143,
+      "step": 18141
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.907845579078455e-05,
+      "loss": 0.3391,
+      "step": 18142
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.906600249066003e-05,
+      "loss": 0.1316,
+      "step": 18143
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.905354919053547e-05,
+      "loss": 0.173,
+      "step": 18144
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.904109589041094e-05,
+      "loss": 0.0996,
+      "step": 18145
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.902864259028642e-05,
+      "loss": 0.2393,
+      "step": 18146
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.901618929016188e-05,
+      "loss": 0.0979,
+      "step": 18147
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.900373599003735e-05,
+      "loss": 0.1919,
+      "step": 18148
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.899128268991282e-05,
+      "loss": 0.2291,
+      "step": 18149
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.897882938978828e-05,
+      "loss": 0.2068,
+      "step": 18150
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.896637608966375e-05,
+      "loss": 0.1307,
+      "step": 18151
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.895392278953923e-05,
+      "loss": 0.13,
+      "step": 18152
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.894146948941468e-05,
+      "loss": 0.1644,
+      "step": 18153
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.892901618929016e-05,
+      "loss": 0.2518,
+      "step": 18154
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.891656288916563e-05,
+      "loss": 0.2125,
+      "step": 18155
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.890410958904109e-05,
+      "loss": 0.1631,
+      "step": 18156
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.889165628891656e-05,
+      "loss": 0.1512,
+      "step": 18157
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.887920298879202e-05,
+      "loss": 0.2115,
+      "step": 18158
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.88667496886675e-05,
+      "loss": 0.1209,
+      "step": 18159
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.885429638854295e-05,
+      "loss": 0.2428,
+      "step": 18160
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.884184308841841e-05,
+      "loss": 0.2707,
+      "step": 18161
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.882938978829388e-05,
+      "loss": 0.1411,
+      "step": 18162
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.881693648816936e-05,
+      "loss": 0.1833,
+      "step": 18163
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.880448318804482e-05,
+      "loss": 0.1304,
+      "step": 18164
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.879202988792029e-05,
+      "loss": 0.1336,
+      "step": 18165
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.877957658779576e-05,
+      "loss": 0.153,
+      "step": 18166
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.876712328767122e-05,
+      "loss": 0.297,
+      "step": 18167
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.87546699875467e-05,
+      "loss": 0.2006,
+      "step": 18168
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.874221668742215e-05,
+      "loss": 0.1407,
+      "step": 18169
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.872976338729763e-05,
+      "loss": 0.1751,
+      "step": 18170
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.87173100871731e-05,
+      "loss": 0.1067,
+      "step": 18171
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.870485678704856e-05,
+      "loss": 0.1134,
+      "step": 18172
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.869240348692403e-05,
+      "loss": 0.2654,
+      "step": 18173
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.86799501867995e-05,
+      "loss": 0.3101,
+      "step": 18174
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.866749688667496e-05,
+      "loss": 0.1689,
+      "step": 18175
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.865504358655042e-05,
+      "loss": 0.258,
+      "step": 18176
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.86425902864259e-05,
+      "loss": 0.1538,
+      "step": 18177
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.863013698630135e-05,
+      "loss": 0.1009,
+      "step": 18178
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.861768368617683e-05,
+      "loss": 0.1924,
+      "step": 18179
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.860523038605229e-05,
+      "loss": 0.1954,
+      "step": 18180
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.859277708592776e-05,
+      "loss": 0.1462,
+      "step": 18181
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.858032378580323e-05,
+      "loss": 0.1272,
+      "step": 18182
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.856787048567869e-05,
+      "loss": 0.154,
+      "step": 18183
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.855541718555416e-05,
+      "loss": 0.1304,
+      "step": 18184
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.854296388542964e-05,
+      "loss": 0.1452,
+      "step": 18185
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.85305105853051e-05,
+      "loss": 0.2763,
+      "step": 18186
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.851805728518057e-05,
+      "loss": 0.1927,
+      "step": 18187
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.850560398505604e-05,
+      "loss": 0.1805,
+      "step": 18188
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.84931506849315e-05,
+      "loss": 0.2566,
+      "step": 18189
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.848069738480697e-05,
+      "loss": 0.1644,
+      "step": 18190
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.846824408468244e-05,
+      "loss": 0.2783,
+      "step": 18191
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.845579078455789e-05,
+      "loss": 0.2863,
+      "step": 18192
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.844333748443336e-05,
+      "loss": 0.2278,
+      "step": 18193
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.843088418430882e-05,
+      "loss": 0.2869,
+      "step": 18194
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.84184308841843e-05,
+      "loss": 0.1454,
+      "step": 18195
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.840597758405977e-05,
+      "loss": 0.1296,
+      "step": 18196
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.839352428393523e-05,
+      "loss": 0.1425,
+      "step": 18197
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.83810709838107e-05,
+      "loss": 0.2046,
+      "step": 18198
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.836861768368617e-05,
+      "loss": 0.1535,
+      "step": 18199
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.835616438356163e-05,
+      "loss": 0.1318,
+      "step": 18200
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.83437110834371e-05,
+      "loss": 0.1555,
+      "step": 18201
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.833125778331258e-05,
+      "loss": 0.2051,
+      "step": 18202
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.831880448318804e-05,
+      "loss": 0.1118,
+      "step": 18203
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.830635118306351e-05,
+      "loss": 0.2477,
+      "step": 18204
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.829389788293898e-05,
+      "loss": 0.155,
+      "step": 18205
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.828144458281444e-05,
+      "loss": 0.249,
+      "step": 18206
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.82689912826899e-05,
+      "loss": 0.1472,
+      "step": 18207
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.825653798256536e-05,
+      "loss": 0.299,
+      "step": 18208
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.824408468244083e-05,
+      "loss": 0.105,
+      "step": 18209
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.82316313823163e-05,
+      "loss": 0.1836,
+      "step": 18210
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.821917808219176e-05,
+      "loss": 0.2387,
+      "step": 18211
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.820672478206724e-05,
+      "loss": 0.1789,
+      "step": 18212
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.819427148194271e-05,
+      "loss": 0.1574,
+      "step": 18213
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.818181818181817e-05,
+      "loss": 0.2773,
+      "step": 18214
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.816936488169364e-05,
+      "loss": 0.1081,
+      "step": 18215
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.815691158156911e-05,
+      "loss": 0.1801,
+      "step": 18216
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.814445828144457e-05,
+      "loss": 0.2708,
+      "step": 18217
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.813200498132005e-05,
+      "loss": 0.14,
+      "step": 18218
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.811955168119552e-05,
+      "loss": 0.181,
+      "step": 18219
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.810709838107098e-05,
+      "loss": 0.1478,
+      "step": 18220
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.809464508094645e-05,
+      "loss": 0.1398,
+      "step": 18221
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.808219178082192e-05,
+      "loss": 0.1128,
+      "step": 18222
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.806973848069737e-05,
+      "loss": 0.2523,
+      "step": 18223
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.805728518057284e-05,
+      "loss": 0.1615,
+      "step": 18224
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.80448318804483e-05,
+      "loss": 0.1508,
+      "step": 18225
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.803237858032377e-05,
+      "loss": 0.1811,
+      "step": 18226
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.801992528019925e-05,
+      "loss": 0.1216,
+      "step": 18227
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.80074719800747e-05,
+      "loss": 0.1858,
+      "step": 18228
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.799501867995018e-05,
+      "loss": 0.2005,
+      "step": 18229
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.798256537982565e-05,
+      "loss": 0.1777,
+      "step": 18230
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.797011207970111e-05,
+      "loss": 0.1647,
+      "step": 18231
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.795765877957658e-05,
+      "loss": 0.1434,
+      "step": 18232
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.794520547945206e-05,
+      "loss": 0.1434,
+      "step": 18233
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.793275217932751e-05,
+      "loss": 0.1692,
+      "step": 18234
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.792029887920299e-05,
+      "loss": 0.135,
+      "step": 18235
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.790784557907846e-05,
+      "loss": 0.1845,
+      "step": 18236
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.789539227895392e-05,
+      "loss": 0.1524,
+      "step": 18237
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.788293897882939e-05,
+      "loss": 0.2261,
+      "step": 18238
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.787048567870484e-05,
+      "loss": 0.2064,
+      "step": 18239
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.785803237858031e-05,
+      "loss": 0.1163,
+      "step": 18240
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.784557907845578e-05,
+      "loss": 0.1206,
+      "step": 18241
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.783312577833124e-05,
+      "loss": 0.2404,
+      "step": 18242
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.782067247820671e-05,
+      "loss": 0.1456,
+      "step": 18243
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.780821917808219e-05,
+      "loss": 0.1363,
+      "step": 18244
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.779576587795765e-05,
+      "loss": 0.1276,
+      "step": 18245
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.778331257783312e-05,
+      "loss": 0.1467,
+      "step": 18246
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.777085927770859e-05,
+      "loss": 0.2023,
+      "step": 18247
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.775840597758405e-05,
+      "loss": 0.2296,
+      "step": 18248
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.774595267745952e-05,
+      "loss": 0.1671,
+      "step": 18249
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.773349937733498e-05,
+      "loss": 0.1375,
+      "step": 18250
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.772104607721046e-05,
+      "loss": 0.1534,
+      "step": 18251
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.770859277708593e-05,
+      "loss": 0.1566,
+      "step": 18252
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.769613947696139e-05,
+      "loss": 0.1371,
+      "step": 18253
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.768368617683685e-05,
+      "loss": 0.229,
+      "step": 18254
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.767123287671232e-05,
+      "loss": 0.1371,
+      "step": 18255
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.765877957658778e-05,
+      "loss": 0.1565,
+      "step": 18256
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.764632627646325e-05,
+      "loss": 0.1464,
+      "step": 18257
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.763387297633872e-05,
+      "loss": 0.1176,
+      "step": 18258
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.762141967621418e-05,
+      "loss": 0.1508,
+      "step": 18259
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.760896637608966e-05,
+      "loss": 0.1282,
+      "step": 18260
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.759651307596512e-05,
+      "loss": 0.284,
+      "step": 18261
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.758405977584059e-05,
+      "loss": 0.1292,
+      "step": 18262
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.757160647571606e-05,
+      "loss": 0.2719,
+      "step": 18263
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.755915317559152e-05,
+      "loss": 0.1872,
+      "step": 18264
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.754669987546699e-05,
+      "loss": 0.1796,
+      "step": 18265
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.753424657534247e-05,
+      "loss": 0.1658,
+      "step": 18266
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.752179327521792e-05,
+      "loss": 0.1497,
+      "step": 18267
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.75093399750934e-05,
+      "loss": 0.093,
+      "step": 18268
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.749688667496887e-05,
+      "loss": 0.1004,
+      "step": 18269
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.748443337484432e-05,
+      "loss": 0.1835,
+      "step": 18270
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.747198007471979e-05,
+      "loss": 0.163,
+      "step": 18271
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.745952677459525e-05,
+      "loss": 0.198,
+      "step": 18272
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.744707347447072e-05,
+      "loss": 0.0905,
+      "step": 18273
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.743462017434619e-05,
+      "loss": 0.2385,
+      "step": 18274
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.742216687422165e-05,
+      "loss": 0.112,
+      "step": 18275
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.740971357409712e-05,
+      "loss": 0.2664,
+      "step": 18276
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.73972602739726e-05,
+      "loss": 0.2402,
+      "step": 18277
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.738480697384806e-05,
+      "loss": 0.2007,
+      "step": 18278
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.737235367372353e-05,
+      "loss": 0.1133,
+      "step": 18279
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.7359900373599e-05,
+      "loss": 0.2068,
+      "step": 18280
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.734744707347446e-05,
+      "loss": 0.1074,
+      "step": 18281
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.733499377334993e-05,
+      "loss": 0.2838,
+      "step": 18282
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.73225404732254e-05,
+      "loss": 0.1596,
+      "step": 18283
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.731008717310087e-05,
+      "loss": 0.1924,
+      "step": 18284
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.729763387297634e-05,
+      "loss": 0.1237,
+      "step": 18285
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.728518057285178e-05,
+      "loss": 0.125,
+      "step": 18286
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.727272727272726e-05,
+      "loss": 0.1463,
+      "step": 18287
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.726027397260273e-05,
+      "loss": 0.1954,
+      "step": 18288
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.724782067247819e-05,
+      "loss": 0.1304,
+      "step": 18289
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.723536737235366e-05,
+      "loss": 0.1325,
+      "step": 18290
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.722291407222913e-05,
+      "loss": 0.1527,
+      "step": 18291
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.72104607721046e-05,
+      "loss": 0.1141,
+      "step": 18292
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.719800747198007e-05,
+      "loss": 0.0885,
+      "step": 18293
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.718555417185554e-05,
+      "loss": 0.0705,
+      "step": 18294
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.7173100871731e-05,
+      "loss": 0.2562,
+      "step": 18295
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.716064757160647e-05,
+      "loss": 0.2721,
+      "step": 18296
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.714819427148194e-05,
+      "loss": 0.1661,
+      "step": 18297
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.71357409713574e-05,
+      "loss": 0.175,
+      "step": 18298
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.712328767123288e-05,
+      "loss": 0.1903,
+      "step": 18299
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.711083437110835e-05,
+      "loss": 0.1734,
+      "step": 18300
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.70983810709838e-05,
+      "loss": 0.2827,
+      "step": 18301
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.708592777085927e-05,
+      "loss": 0.1414,
+      "step": 18302
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.707347447073473e-05,
+      "loss": 0.1551,
+      "step": 18303
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.70610211706102e-05,
+      "loss": 0.0719,
+      "step": 18304
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.704856787048567e-05,
+      "loss": 0.152,
+      "step": 18305
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.703611457036113e-05,
+      "loss": 0.1479,
+      "step": 18306
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.70236612702366e-05,
+      "loss": 0.2044,
+      "step": 18307
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.701120797011208e-05,
+      "loss": 0.1298,
+      "step": 18308
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.699875466998753e-05,
+      "loss": 0.2498,
+      "step": 18309
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.698630136986301e-05,
+      "loss": 0.1363,
+      "step": 18310
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.697384806973848e-05,
+      "loss": 0.134,
+      "step": 18311
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.696139476961394e-05,
+      "loss": 0.1257,
+      "step": 18312
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.694894146948941e-05,
+      "loss": 0.2241,
+      "step": 18313
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.693648816936488e-05,
+      "loss": 0.1749,
+      "step": 18314
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.692403486924034e-05,
+      "loss": 0.1918,
+      "step": 18315
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.691158156911582e-05,
+      "loss": 0.096,
+      "step": 18316
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.689912826899126e-05,
+      "loss": 0.1131,
+      "step": 18317
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.688667496886673e-05,
+      "loss": 0.1304,
+      "step": 18318
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.687422166874221e-05,
+      "loss": 0.146,
+      "step": 18319
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.686176836861767e-05,
+      "loss": 0.3562,
+      "step": 18320
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.684931506849314e-05,
+      "loss": 0.1183,
+      "step": 18321
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.683686176836861e-05,
+      "loss": 0.0991,
+      "step": 18322
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.682440846824407e-05,
+      "loss": 0.2077,
+      "step": 18323
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.681195516811954e-05,
+      "loss": 0.1643,
+      "step": 18324
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.679950186799502e-05,
+      "loss": 0.1169,
+      "step": 18325
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.678704856787048e-05,
+      "loss": 0.2306,
+      "step": 18326
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.677459526774595e-05,
+      "loss": 0.1958,
+      "step": 18327
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.676214196762142e-05,
+      "loss": 0.1632,
+      "step": 18328
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.674968866749688e-05,
+      "loss": 0.0968,
+      "step": 18329
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.673723536737235e-05,
+      "loss": 0.1136,
+      "step": 18330
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.672478206724781e-05,
+      "loss": 0.0935,
+      "step": 18331
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.671232876712329e-05,
+      "loss": 0.1418,
+      "step": 18332
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.669987546699874e-05,
+      "loss": 0.1618,
+      "step": 18333
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.66874221668742e-05,
+      "loss": 0.1457,
+      "step": 18334
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.667496886674968e-05,
+      "loss": 0.1823,
+      "step": 18335
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.666251556662515e-05,
+      "loss": 0.0978,
+      "step": 18336
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.665006226650061e-05,
+      "loss": 0.1692,
+      "step": 18337
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.663760896637608e-05,
+      "loss": 0.2812,
+      "step": 18338
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.662515566625154e-05,
+      "loss": 0.2231,
+      "step": 18339
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.661270236612701e-05,
+      "loss": 0.1579,
+      "step": 18340
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.660024906600249e-05,
+      "loss": 0.161,
+      "step": 18341
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.658779576587794e-05,
+      "loss": 0.0999,
+      "step": 18342
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.657534246575342e-05,
+      "loss": 0.1958,
+      "step": 18343
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.656288916562889e-05,
+      "loss": 0.1464,
+      "step": 18344
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.655043586550435e-05,
+      "loss": 0.2988,
+      "step": 18345
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.653798256537982e-05,
+      "loss": 0.1986,
+      "step": 18346
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.65255292652553e-05,
+      "loss": 0.1234,
+      "step": 18347
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.651307596513075e-05,
+      "loss": 0.1797,
+      "step": 18348
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.650062266500621e-05,
+      "loss": 0.1058,
+      "step": 18349
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.648816936488167e-05,
+      "loss": 0.0982,
+      "step": 18350
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.647571606475715e-05,
+      "loss": 0.2966,
+      "step": 18351
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.646326276463262e-05,
+      "loss": 0.1655,
+      "step": 18352
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.645080946450808e-05,
+      "loss": 0.1171,
+      "step": 18353
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.643835616438355e-05,
+      "loss": 0.1676,
+      "step": 18354
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.642590286425902e-05,
+      "loss": 0.1032,
+      "step": 18355
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.641344956413448e-05,
+      "loss": 0.1107,
+      "step": 18356
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.640099626400995e-05,
+      "loss": 0.1637,
+      "step": 18357
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.638854296388543e-05,
+      "loss": 0.1921,
+      "step": 18358
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.637608966376089e-05,
+      "loss": 0.1042,
+      "step": 18359
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.636363636363636e-05,
+      "loss": 0.1751,
+      "step": 18360
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.635118306351183e-05,
+      "loss": 0.0737,
+      "step": 18361
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.633872976338729e-05,
+      "loss": 0.0882,
+      "step": 18362
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.632627646326276e-05,
+      "loss": 0.1998,
+      "step": 18363
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.631382316313821e-05,
+      "loss": 0.1874,
+      "step": 18364
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.630136986301368e-05,
+      "loss": 0.179,
+      "step": 18365
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.628891656288915e-05,
+      "loss": 0.1649,
+      "step": 18366
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.627646326276461e-05,
+      "loss": 0.1722,
+      "step": 18367
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.626400996264009e-05,
+      "loss": 0.146,
+      "step": 18368
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.625155666251556e-05,
+      "loss": 0.1307,
+      "step": 18369
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.623910336239102e-05,
+      "loss": 0.2348,
+      "step": 18370
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.622665006226649e-05,
+      "loss": 0.1807,
+      "step": 18371
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.621419676214196e-05,
+      "loss": 0.1567,
+      "step": 18372
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.620174346201742e-05,
+      "loss": 0.1793,
+      "step": 18373
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.61892901618929e-05,
+      "loss": 0.2027,
+      "step": 18374
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.617683686176837e-05,
+      "loss": 0.1152,
+      "step": 18375
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.616438356164383e-05,
+      "loss": 0.207,
+      "step": 18376
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.61519302615193e-05,
+      "loss": 0.1315,
+      "step": 18377
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.613947696139477e-05,
+      "loss": 0.2226,
+      "step": 18378
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.612702366127023e-05,
+      "loss": 0.1345,
+      "step": 18379
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.611457036114569e-05,
+      "loss": 0.1018,
+      "step": 18380
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.610211706102115e-05,
+      "loss": 0.1014,
+      "step": 18381
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.608966376089662e-05,
+      "loss": 0.2481,
+      "step": 18382
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.60772104607721e-05,
+      "loss": 0.1879,
+      "step": 18383
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.606475716064756e-05,
+      "loss": 0.12,
+      "step": 18384
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.605230386052303e-05,
+      "loss": 0.1852,
+      "step": 18385
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.60398505603985e-05,
+      "loss": 0.1547,
+      "step": 18386
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.602739726027396e-05,
+      "loss": 0.1454,
+      "step": 18387
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.601494396014943e-05,
+      "loss": 0.1376,
+      "step": 18388
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.60024906600249e-05,
+      "loss": 0.1874,
+      "step": 18389
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.599003735990036e-05,
+      "loss": 0.1645,
+      "step": 18390
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.597758405977584e-05,
+      "loss": 0.2094,
+      "step": 18391
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.596513075965131e-05,
+      "loss": 0.14,
+      "step": 18392
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.595267745952677e-05,
+      "loss": 0.1577,
+      "step": 18393
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.594022415940224e-05,
+      "loss": 0.2281,
+      "step": 18394
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.592777085927771e-05,
+      "loss": 0.2302,
+      "step": 18395
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.591531755915316e-05,
+      "loss": 0.1105,
+      "step": 18396
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.590286425902863e-05,
+      "loss": 0.1733,
+      "step": 18397
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.589041095890409e-05,
+      "loss": 0.1346,
+      "step": 18398
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.587795765877956e-05,
+      "loss": 0.1531,
+      "step": 18399
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.586550435865504e-05,
+      "loss": 0.1637,
+      "step": 18400
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.58530510585305e-05,
+      "loss": 0.2593,
+      "step": 18401
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.584059775840597e-05,
+      "loss": 0.1782,
+      "step": 18402
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.582814445828144e-05,
+      "loss": 0.1353,
+      "step": 18403
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.58156911581569e-05,
+      "loss": 0.076,
+      "step": 18404
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.580323785803237e-05,
+      "loss": 0.1505,
+      "step": 18405
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.579078455790785e-05,
+      "loss": 0.1312,
+      "step": 18406
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.57783312577833e-05,
+      "loss": 0.1484,
+      "step": 18407
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.576587795765878e-05,
+      "loss": 0.1855,
+      "step": 18408
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.575342465753425e-05,
+      "loss": 0.2012,
+      "step": 18409
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.574097135740971e-05,
+      "loss": 0.1823,
+      "step": 18410
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.572851805728517e-05,
+      "loss": 0.1104,
+      "step": 18411
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.571606475716063e-05,
+      "loss": 0.0812,
+      "step": 18412
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.57036114570361e-05,
+      "loss": 0.2013,
+      "step": 18413
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.569115815691157e-05,
+      "loss": 0.2014,
+      "step": 18414
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.567870485678703e-05,
+      "loss": 0.1343,
+      "step": 18415
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.56662515566625e-05,
+      "loss": 0.1225,
+      "step": 18416
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.565379825653798e-05,
+      "loss": 0.152,
+      "step": 18417
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.564134495641344e-05,
+      "loss": 0.1675,
+      "step": 18418
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.562889165628891e-05,
+      "loss": 0.1295,
+      "step": 18419
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.561643835616437e-05,
+      "loss": 0.1875,
+      "step": 18420
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.560398505603984e-05,
+      "loss": 0.2735,
+      "step": 18421
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.559153175591532e-05,
+      "loss": 0.1144,
+      "step": 18422
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.557907845579077e-05,
+      "loss": 0.1246,
+      "step": 18423
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.556662515566625e-05,
+      "loss": 0.1162,
+      "step": 18424
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.555417185554172e-05,
+      "loss": 0.117,
+      "step": 18425
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.554171855541718e-05,
+      "loss": 0.2227,
+      "step": 18426
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.552926525529264e-05,
+      "loss": 0.1467,
+      "step": 18427
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.551681195516811e-05,
+      "loss": 0.2042,
+      "step": 18428
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.550435865504357e-05,
+      "loss": 0.128,
+      "step": 18429
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.549190535491904e-05,
+      "loss": 0.1367,
+      "step": 18430
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.54794520547945e-05,
+      "loss": 0.1563,
+      "step": 18431
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.546699875466997e-05,
+      "loss": 0.3719,
+      "step": 18432
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.545454545454545e-05,
+      "loss": 0.2716,
+      "step": 18433
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.54420921544209e-05,
+      "loss": 0.1437,
+      "step": 18434
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.542963885429638e-05,
+      "loss": 0.161,
+      "step": 18435
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.541718555417185e-05,
+      "loss": 0.0789,
+      "step": 18436
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.540473225404731e-05,
+      "loss": 0.1712,
+      "step": 18437
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.539227895392278e-05,
+      "loss": 0.2547,
+      "step": 18438
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.537982565379826e-05,
+      "loss": 0.2132,
+      "step": 18439
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.536737235367372e-05,
+      "loss": 0.2121,
+      "step": 18440
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.535491905354919e-05,
+      "loss": 0.1312,
+      "step": 18441
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.534246575342466e-05,
+      "loss": 0.1413,
+      "step": 18442
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.533001245330011e-05,
+      "loss": 0.1023,
+      "step": 18443
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.531755915317558e-05,
+      "loss": 0.1607,
+      "step": 18444
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.530510585305104e-05,
+      "loss": 0.208,
+      "step": 18445
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.529265255292651e-05,
+      "loss": 0.2141,
+      "step": 18446
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.528019925280198e-05,
+      "loss": 0.1663,
+      "step": 18447
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.526774595267744e-05,
+      "loss": 0.2242,
+      "step": 18448
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.525529265255292e-05,
+      "loss": 0.0955,
+      "step": 18449
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.524283935242839e-05,
+      "loss": 0.1098,
+      "step": 18450
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.523038605230385e-05,
+      "loss": 0.2391,
+      "step": 18451
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.521793275217932e-05,
+      "loss": 0.1787,
+      "step": 18452
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.52054794520548e-05,
+      "loss": 0.1799,
+      "step": 18453
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.519302615193025e-05,
+      "loss": 0.0968,
+      "step": 18454
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.518057285180573e-05,
+      "loss": 0.1286,
+      "step": 18455
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.51681195516812e-05,
+      "loss": 0.2019,
+      "step": 18456
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.515566625155666e-05,
+      "loss": 0.2782,
+      "step": 18457
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.514321295143212e-05,
+      "loss": 0.1848,
+      "step": 18458
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.513075965130758e-05,
+      "loss": 0.1536,
+      "step": 18459
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.511830635118305e-05,
+      "loss": 0.1709,
+      "step": 18460
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.510585305105852e-05,
+      "loss": 0.1066,
+      "step": 18461
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.509339975093398e-05,
+      "loss": 0.0978,
+      "step": 18462
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.508094645080945e-05,
+      "loss": 0.244,
+      "step": 18463
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.506849315068493e-05,
+      "loss": 0.1743,
+      "step": 18464
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.505603985056038e-05,
+      "loss": 0.1151,
+      "step": 18465
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.504358655043586e-05,
+      "loss": 0.1348,
+      "step": 18466
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.503113325031133e-05,
+      "loss": 0.0922,
+      "step": 18467
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.501867995018679e-05,
+      "loss": 0.1157,
+      "step": 18468
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.500622665006226e-05,
+      "loss": 0.194,
+      "step": 18469
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.499377334993773e-05,
+      "loss": 0.2161,
+      "step": 18470
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.49813200498132e-05,
+      "loss": 0.2083,
+      "step": 18471
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.496886674968867e-05,
+      "loss": 0.1312,
+      "step": 18472
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.495641344956414e-05,
+      "loss": 0.1213,
+      "step": 18473
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.494396014943959e-05,
+      "loss": 0.1203,
+      "step": 18474
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.493150684931506e-05,
+      "loss": 0.1081,
+      "step": 18475
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.491905354919052e-05,
+      "loss": 0.2023,
+      "step": 18476
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.490660024906599e-05,
+      "loss": 0.1925,
+      "step": 18477
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.489414694894146e-05,
+      "loss": 0.1218,
+      "step": 18478
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.488169364881692e-05,
+      "loss": 0.1879,
+      "step": 18479
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.48692403486924e-05,
+      "loss": 0.1982,
+      "step": 18480
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.485678704856787e-05,
+      "loss": 0.1059,
+      "step": 18481
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.484433374844333e-05,
+      "loss": 0.2685,
+      "step": 18482
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.48318804483188e-05,
+      "loss": 0.1339,
+      "step": 18483
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.481942714819427e-05,
+      "loss": 0.2257,
+      "step": 18484
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.480697384806973e-05,
+      "loss": 0.1536,
+      "step": 18485
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.47945205479452e-05,
+      "loss": 0.1342,
+      "step": 18486
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.478206724782068e-05,
+      "loss": 0.1243,
+      "step": 18487
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.476961394769614e-05,
+      "loss": 0.3277,
+      "step": 18488
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.475716064757161e-05,
+      "loss": 0.1817,
+      "step": 18489
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.474470734744705e-05,
+      "loss": 0.1823,
+      "step": 18490
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.473225404732253e-05,
+      "loss": 0.2262,
+      "step": 18491
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.4719800747198e-05,
+      "loss": 0.1137,
+      "step": 18492
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.470734744707346e-05,
+      "loss": 0.1869,
+      "step": 18493
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.469489414694893e-05,
+      "loss": 0.1509,
+      "step": 18494
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.46824408468244e-05,
+      "loss": 0.2659,
+      "step": 18495
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.466998754669986e-05,
+      "loss": 0.1726,
+      "step": 18496
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.465753424657534e-05,
+      "loss": 0.2408,
+      "step": 18497
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.464508094645081e-05,
+      "loss": 0.1715,
+      "step": 18498
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.463262764632627e-05,
+      "loss": 0.1523,
+      "step": 18499
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.462017434620174e-05,
+      "loss": 0.1487,
+      "step": 18500
+    },
+    {
+      "epoch": 7.09,
+      "eval_cer": 0.10297130148077335,
+      "eval_loss": 0.3816582262516022,
+      "eval_runtime": 47.8664,
+      "eval_samples_per_second": 36.623,
+      "eval_steps_per_second": 4.596,
+      "eval_wer": 0.16712675928859122,
+      "step": 18500
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.46077210460772e-05,
+      "loss": 0.202,
+      "step": 18501
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.459526774595267e-05,
+      "loss": 0.1241,
+      "step": 18502
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.458281444582814e-05,
+      "loss": 0.1252,
+      "step": 18503
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.45703611457036e-05,
+      "loss": 0.1297,
+      "step": 18504
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.455790784557908e-05,
+      "loss": 0.1539,
+      "step": 18505
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.454545454545454e-05,
+      "loss": 0.1323,
+      "step": 18506
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.453300124533e-05,
+      "loss": 0.1818,
+      "step": 18507
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.452054794520547e-05,
+      "loss": 0.1655,
+      "step": 18508
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.450809464508094e-05,
+      "loss": 0.111,
+      "step": 18509
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.44956413449564e-05,
+      "loss": 0.1207,
+      "step": 18510
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.448318804483187e-05,
+      "loss": 0.0727,
+      "step": 18511
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.447073474470733e-05,
+      "loss": 0.1323,
+      "step": 18512
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.44582814445828e-05,
+      "loss": 0.2054,
+      "step": 18513
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.444582814445828e-05,
+      "loss": 0.1587,
+      "step": 18514
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.443337484433374e-05,
+      "loss": 0.131,
+      "step": 18515
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.442092154420921e-05,
+      "loss": 0.2028,
+      "step": 18516
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.440846824408468e-05,
+      "loss": 0.0912,
+      "step": 18517
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.439601494396014e-05,
+      "loss": 0.1673,
+      "step": 18518
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.438356164383561e-05,
+      "loss": 0.1155,
+      "step": 18519
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.437110834371109e-05,
+      "loss": 0.2369,
+      "step": 18520
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.435865504358653e-05,
+      "loss": 0.2025,
+      "step": 18521
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.4346201743462e-05,
+      "loss": 0.2363,
+      "step": 18522
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.433374844333746e-05,
+      "loss": 0.1234,
+      "step": 18523
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.432129514321294e-05,
+      "loss": 0.1819,
+      "step": 18524
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.430884184308841e-05,
+      "loss": 0.085,
+      "step": 18525
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.429638854296387e-05,
+      "loss": 0.3237,
+      "step": 18526
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.428393524283934e-05,
+      "loss": 0.1926,
+      "step": 18527
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.427148194271481e-05,
+      "loss": 0.1614,
+      "step": 18528
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.425902864259027e-05,
+      "loss": 0.0839,
+      "step": 18529
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.424657534246575e-05,
+      "loss": 0.1017,
+      "step": 18530
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.423412204234122e-05,
+      "loss": 0.1067,
+      "step": 18531
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.422166874221668e-05,
+      "loss": 0.2187,
+      "step": 18532
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.420921544209215e-05,
+      "loss": 0.2168,
+      "step": 18533
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.419676214196762e-05,
+      "loss": 0.1736,
+      "step": 18534
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.418430884184308e-05,
+      "loss": 0.1237,
+      "step": 18535
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.417185554171855e-05,
+      "loss": 0.1525,
+      "step": 18536
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.4159402241594e-05,
+      "loss": 0.1898,
+      "step": 18537
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.414694894146947e-05,
+      "loss": 0.2427,
+      "step": 18538
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.413449564134495e-05,
+      "loss": 0.1424,
+      "step": 18539
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.41220423412204e-05,
+      "loss": 0.1029,
+      "step": 18540
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.410958904109588e-05,
+      "loss": 0.1256,
+      "step": 18541
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.409713574097135e-05,
+      "loss": 0.1812,
+      "step": 18542
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.408468244084681e-05,
+      "loss": 0.138,
+      "step": 18543
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.407222914072228e-05,
+      "loss": 0.1932,
+      "step": 18544
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.405977584059776e-05,
+      "loss": 0.274,
+      "step": 18545
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.404732254047321e-05,
+      "loss": 0.3105,
+      "step": 18546
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.403486924034869e-05,
+      "loss": 0.1656,
+      "step": 18547
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.402241594022416e-05,
+      "loss": 0.0995,
+      "step": 18548
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.400996264009962e-05,
+      "loss": 0.1084,
+      "step": 18549
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.399750933997509e-05,
+      "loss": 0.1042,
+      "step": 18550
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.398505603985056e-05,
+      "loss": 0.1901,
+      "step": 18551
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.397260273972602e-05,
+      "loss": 0.1925,
+      "step": 18552
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.396014943960148e-05,
+      "loss": 0.2094,
+      "step": 18553
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.394769613947694e-05,
+      "loss": 0.0949,
+      "step": 18554
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.393524283935241e-05,
+      "loss": 0.1136,
+      "step": 18555
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.392278953922789e-05,
+      "loss": 0.1115,
+      "step": 18556
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.391033623910335e-05,
+      "loss": 0.1708,
+      "step": 18557
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.389788293897882e-05,
+      "loss": 0.2179,
+      "step": 18558
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.388542963885429e-05,
+      "loss": 0.1447,
+      "step": 18559
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.387297633872975e-05,
+      "loss": 0.1539,
+      "step": 18560
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.386052303860522e-05,
+      "loss": 0.1629,
+      "step": 18561
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.38480697384807e-05,
+      "loss": 0.1542,
+      "step": 18562
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.383561643835616e-05,
+      "loss": 0.1822,
+      "step": 18563
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.382316313823163e-05,
+      "loss": 0.1543,
+      "step": 18564
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.38107098381071e-05,
+      "loss": 0.1011,
+      "step": 18565
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.379825653798256e-05,
+      "loss": 0.1355,
+      "step": 18566
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.378580323785803e-05,
+      "loss": 0.1993,
+      "step": 18567
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.377334993773348e-05,
+      "loss": 0.1511,
+      "step": 18568
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.376089663760895e-05,
+      "loss": 0.1784,
+      "step": 18569
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.374844333748442e-05,
+      "loss": 0.15,
+      "step": 18570
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.373599003735988e-05,
+      "loss": 0.1422,
+      "step": 18571
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.372353673723536e-05,
+      "loss": 0.1544,
+      "step": 18572
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.371108343711083e-05,
+      "loss": 0.242,
+      "step": 18573
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.369863013698629e-05,
+      "loss": 0.1185,
+      "step": 18574
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.368617683686176e-05,
+      "loss": 0.119,
+      "step": 18575
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.367372353673723e-05,
+      "loss": 0.1818,
+      "step": 18576
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.366127023661269e-05,
+      "loss": 0.2801,
+      "step": 18577
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.364881693648817e-05,
+      "loss": 0.1368,
+      "step": 18578
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.363636363636364e-05,
+      "loss": 0.1437,
+      "step": 18579
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.36239103362391e-05,
+      "loss": 0.1436,
+      "step": 18580
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.361145703611457e-05,
+      "loss": 0.1163,
+      "step": 18581
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.359900373599003e-05,
+      "loss": 0.2079,
+      "step": 18582
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.35865504358655e-05,
+      "loss": 0.1408,
+      "step": 18583
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.357409713574096e-05,
+      "loss": 0.2595,
+      "step": 18584
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.356164383561642e-05,
+      "loss": 0.1206,
+      "step": 18585
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.354919053549189e-05,
+      "loss": 0.1406,
+      "step": 18586
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.353673723536737e-05,
+      "loss": 0.1358,
+      "step": 18587
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.352428393524282e-05,
+      "loss": 0.1837,
+      "step": 18588
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.35118306351183e-05,
+      "loss": 0.1646,
+      "step": 18589
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.349937733499377e-05,
+      "loss": 0.133,
+      "step": 18590
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.348692403486923e-05,
+      "loss": 0.117,
+      "step": 18591
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.34744707347447e-05,
+      "loss": 0.1249,
+      "step": 18592
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.346201743462016e-05,
+      "loss": 0.1152,
+      "step": 18593
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.344956413449563e-05,
+      "loss": 0.1325,
+      "step": 18594
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.34371108343711e-05,
+      "loss": 0.2481,
+      "step": 18595
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.342465753424657e-05,
+      "loss": 0.1502,
+      "step": 18596
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.341220423412204e-05,
+      "loss": 0.1843,
+      "step": 18597
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.339975093399751e-05,
+      "loss": 0.0925,
+      "step": 18598
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.338729763387297e-05,
+      "loss": 0.0962,
+      "step": 18599
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.337484433374843e-05,
+      "loss": 0.2286,
+      "step": 18600
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.33623910336239e-05,
+      "loss": 0.2579,
+      "step": 18601
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.334993773349936e-05,
+      "loss": 0.1679,
+      "step": 18602
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.333748443337483e-05,
+      "loss": 0.2041,
+      "step": 18603
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.33250311332503e-05,
+      "loss": 0.1117,
+      "step": 18604
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.331257783312577e-05,
+      "loss": 0.1099,
+      "step": 18605
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.330012453300124e-05,
+      "loss": 0.1573,
+      "step": 18606
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.32876712328767e-05,
+      "loss": 0.2642,
+      "step": 18607
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.327521793275217e-05,
+      "loss": 0.127,
+      "step": 18608
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.326276463262764e-05,
+      "loss": 0.2329,
+      "step": 18609
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.32503113325031e-05,
+      "loss": 0.1052,
+      "step": 18610
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.323785803237858e-05,
+      "loss": 0.1611,
+      "step": 18611
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.322540473225405e-05,
+      "loss": 0.1035,
+      "step": 18612
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.321295143212951e-05,
+      "loss": 0.2089,
+      "step": 18613
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.320049813200498e-05,
+      "loss": 0.2777,
+      "step": 18614
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.318804483188045e-05,
+      "loss": 0.1932,
+      "step": 18615
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.31755915317559e-05,
+      "loss": 0.1593,
+      "step": 18616
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.316313823163137e-05,
+      "loss": 0.1653,
+      "step": 18617
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.315068493150683e-05,
+      "loss": 0.1456,
+      "step": 18618
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.31382316313823e-05,
+      "loss": 0.182,
+      "step": 18619
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.312577833125778e-05,
+      "loss": 0.2189,
+      "step": 18620
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.311332503113323e-05,
+      "loss": 0.1565,
+      "step": 18621
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.310087173100871e-05,
+      "loss": 0.3397,
+      "step": 18622
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.308841843088418e-05,
+      "loss": 0.1548,
+      "step": 18623
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.307596513075964e-05,
+      "loss": 0.1293,
+      "step": 18624
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.306351183063511e-05,
+      "loss": 0.1154,
+      "step": 18625
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.305105853051058e-05,
+      "loss": 0.221,
+      "step": 18626
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.303860523038604e-05,
+      "loss": 0.2044,
+      "step": 18627
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.302615193026152e-05,
+      "loss": 0.1722,
+      "step": 18628
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.301369863013699e-05,
+      "loss": 0.1244,
+      "step": 18629
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.300124533001245e-05,
+      "loss": 0.1457,
+      "step": 18630
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.298879202988791e-05,
+      "loss": 0.1792,
+      "step": 18631
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.297633872976337e-05,
+      "loss": 0.1449,
+      "step": 18632
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.296388542963884e-05,
+      "loss": 0.148,
+      "step": 18633
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.295143212951431e-05,
+      "loss": 0.2138,
+      "step": 18634
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.293897882938977e-05,
+      "loss": 0.218,
+      "step": 18635
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.292652552926524e-05,
+      "loss": 0.2107,
+      "step": 18636
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.291407222914072e-05,
+      "loss": 0.1407,
+      "step": 18637
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.290161892901618e-05,
+      "loss": 0.2159,
+      "step": 18638
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.288916562889165e-05,
+      "loss": 0.2366,
+      "step": 18639
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.287671232876712e-05,
+      "loss": 0.1095,
+      "step": 18640
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.286425902864258e-05,
+      "loss": 0.0977,
+      "step": 18641
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.285180572851805e-05,
+      "loss": 0.1025,
+      "step": 18642
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.283935242839353e-05,
+      "loss": 0.1313,
+      "step": 18643
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.282689912826899e-05,
+      "loss": 0.1323,
+      "step": 18644
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.281444582814446e-05,
+      "loss": 0.1623,
+      "step": 18645
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.280199252801993e-05,
+      "loss": 0.1619,
+      "step": 18646
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.278953922789538e-05,
+      "loss": 0.1163,
+      "step": 18647
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.277708592777085e-05,
+      "loss": 0.0784,
+      "step": 18648
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.276463262764631e-05,
+      "loss": 0.1663,
+      "step": 18649
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.275217932752178e-05,
+      "loss": 0.1502,
+      "step": 18650
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.273972602739725e-05,
+      "loss": 0.1842,
+      "step": 18651
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.272727272727271e-05,
+      "loss": 0.1858,
+      "step": 18652
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.271481942714819e-05,
+      "loss": 0.1633,
+      "step": 18653
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.270236612702366e-05,
+      "loss": 0.1402,
+      "step": 18654
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.268991282689912e-05,
+      "loss": 0.1337,
+      "step": 18655
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.267745952677459e-05,
+      "loss": 0.1196,
+      "step": 18656
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.266500622665006e-05,
+      "loss": 0.2664,
+      "step": 18657
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.265255292652552e-05,
+      "loss": 0.1191,
+      "step": 18658
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.2640099626401e-05,
+      "loss": 0.1923,
+      "step": 18659
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.262764632627647e-05,
+      "loss": 0.1121,
+      "step": 18660
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.261519302615193e-05,
+      "loss": 0.2105,
+      "step": 18661
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.26027397260274e-05,
+      "loss": 0.0995,
+      "step": 18662
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.259028642590285e-05,
+      "loss": 0.1357,
+      "step": 18663
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.257783312577832e-05,
+      "loss": 0.2388,
+      "step": 18664
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.256537982565379e-05,
+      "loss": 0.1195,
+      "step": 18665
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.255292652552925e-05,
+      "loss": 0.18,
+      "step": 18666
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.254047322540472e-05,
+      "loss": 0.1346,
+      "step": 18667
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.25280199252802e-05,
+      "loss": 0.1135,
+      "step": 18668
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.251556662515565e-05,
+      "loss": 0.1788,
+      "step": 18669
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.250311332503113e-05,
+      "loss": 0.2505,
+      "step": 18670
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.24906600249066e-05,
+      "loss": 0.1656,
+      "step": 18671
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.247820672478206e-05,
+      "loss": 0.1327,
+      "step": 18672
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.246575342465753e-05,
+      "loss": 0.1287,
+      "step": 18673
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.245330012453299e-05,
+      "loss": 0.1709,
+      "step": 18674
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.244084682440846e-05,
+      "loss": 0.1436,
+      "step": 18675
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.242839352428394e-05,
+      "loss": 0.2883,
+      "step": 18676
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.24159402241594e-05,
+      "loss": 0.1352,
+      "step": 18677
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.240348692403485e-05,
+      "loss": 0.1292,
+      "step": 18678
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.239103362391033e-05,
+      "loss": 0.1054,
+      "step": 18679
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.237858032378579e-05,
+      "loss": 0.1697,
+      "step": 18680
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.236612702366126e-05,
+      "loss": 0.1874,
+      "step": 18681
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.235367372353673e-05,
+      "loss": 0.1513,
+      "step": 18682
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.234122042341219e-05,
+      "loss": 0.16,
+      "step": 18683
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.232876712328766e-05,
+      "loss": 0.2312,
+      "step": 18684
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.231631382316312e-05,
+      "loss": 0.1873,
+      "step": 18685
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.23038605230386e-05,
+      "loss": 0.1207,
+      "step": 18686
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.229140722291407e-05,
+      "loss": 0.1005,
+      "step": 18687
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.227895392278953e-05,
+      "loss": 0.2817,
+      "step": 18688
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.2266500622665e-05,
+      "loss": 0.2519,
+      "step": 18689
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.225404732254047e-05,
+      "loss": 0.1919,
+      "step": 18690
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.224159402241593e-05,
+      "loss": 0.1085,
+      "step": 18691
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.22291407222914e-05,
+      "loss": 0.1282,
+      "step": 18692
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.221668742216688e-05,
+      "loss": 0.1326,
+      "step": 18693
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.220423412204232e-05,
+      "loss": 0.1233,
+      "step": 18694
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.21917808219178e-05,
+      "loss": 0.1934,
+      "step": 18695
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.217932752179326e-05,
+      "loss": 0.2174,
+      "step": 18696
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.216687422166873e-05,
+      "loss": 0.1342,
+      "step": 18697
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.21544209215442e-05,
+      "loss": 0.2649,
+      "step": 18698
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.214196762141966e-05,
+      "loss": 0.1359,
+      "step": 18699
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.212951432129513e-05,
+      "loss": 0.1043,
+      "step": 18700
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.21170610211706e-05,
+      "loss": 0.2452,
+      "step": 18701
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.210460772104606e-05,
+      "loss": 0.1063,
+      "step": 18702
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.209215442092154e-05,
+      "loss": 0.0858,
+      "step": 18703
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.207970112079701e-05,
+      "loss": 0.1912,
+      "step": 18704
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.206724782067247e-05,
+      "loss": 0.1596,
+      "step": 18705
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.205479452054794e-05,
+      "loss": 0.1043,
+      "step": 18706
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.204234122042341e-05,
+      "loss": 0.2166,
+      "step": 18707
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.202988792029887e-05,
+      "loss": 0.1968,
+      "step": 18708
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.201743462017435e-05,
+      "loss": 0.1959,
+      "step": 18709
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.200498132004979e-05,
+      "loss": 0.1623,
+      "step": 18710
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.199252801992526e-05,
+      "loss": 0.1595,
+      "step": 18711
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.198007471980074e-05,
+      "loss": 0.1523,
+      "step": 18712
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.19676214196762e-05,
+      "loss": 0.32,
+      "step": 18713
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.195516811955167e-05,
+      "loss": 0.1548,
+      "step": 18714
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.194271481942714e-05,
+      "loss": 0.1487,
+      "step": 18715
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.19302615193026e-05,
+      "loss": 0.1259,
+      "step": 18716
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.191780821917807e-05,
+      "loss": 0.1281,
+      "step": 18717
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.190535491905355e-05,
+      "loss": 0.1578,
+      "step": 18718
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.1892901618929e-05,
+      "loss": 0.0914,
+      "step": 18719
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.188044831880448e-05,
+      "loss": 0.2787,
+      "step": 18720
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.186799501867995e-05,
+      "loss": 0.1619,
+      "step": 18721
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.185554171855541e-05,
+      "loss": 0.1104,
+      "step": 18722
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.184308841843088e-05,
+      "loss": 0.1341,
+      "step": 18723
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.183063511830636e-05,
+      "loss": 0.1082,
+      "step": 18724
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.18181818181818e-05,
+      "loss": 0.1805,
+      "step": 18725
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.180572851805727e-05,
+      "loss": 0.15,
+      "step": 18726
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.179327521793273e-05,
+      "loss": 0.1859,
+      "step": 18727
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.17808219178082e-05,
+      "loss": 0.2063,
+      "step": 18728
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.176836861768368e-05,
+      "loss": 0.2257,
+      "step": 18729
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.175591531755914e-05,
+      "loss": 0.1713,
+      "step": 18730
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.174346201743461e-05,
+      "loss": 0.1423,
+      "step": 18731
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.173100871731008e-05,
+      "loss": 0.2809,
+      "step": 18732
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.171855541718554e-05,
+      "loss": 0.1797,
+      "step": 18733
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.170610211706102e-05,
+      "loss": 0.1599,
+      "step": 18734
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.169364881693649e-05,
+      "loss": 0.1186,
+      "step": 18735
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.168119551681195e-05,
+      "loss": 0.1626,
+      "step": 18736
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.166874221668742e-05,
+      "loss": 0.111,
+      "step": 18737
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.165628891656289e-05,
+      "loss": 0.2119,
+      "step": 18738
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.164383561643835e-05,
+      "loss": 0.1442,
+      "step": 18739
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.163138231631382e-05,
+      "loss": 0.1613,
+      "step": 18740
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.161892901618927e-05,
+      "loss": 0.1912,
+      "step": 18741
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.160647571606474e-05,
+      "loss": 0.169,
+      "step": 18742
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.159402241594022e-05,
+      "loss": 0.0942,
+      "step": 18743
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.158156911581567e-05,
+      "loss": 0.1426,
+      "step": 18744
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.156911581569115e-05,
+      "loss": 0.2631,
+      "step": 18745
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.155666251556662e-05,
+      "loss": 0.1474,
+      "step": 18746
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.154420921544208e-05,
+      "loss": 0.1416,
+      "step": 18747
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.153175591531755e-05,
+      "loss": 0.1217,
+      "step": 18748
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.151930261519302e-05,
+      "loss": 0.144,
+      "step": 18749
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.150684931506848e-05,
+      "loss": 0.1291,
+      "step": 18750
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.149439601494396e-05,
+      "loss": 0.1679,
+      "step": 18751
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.148194271481943e-05,
+      "loss": 0.1398,
+      "step": 18752
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.146948941469489e-05,
+      "loss": 0.0917,
+      "step": 18753
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.145703611457036e-05,
+      "loss": 0.1702,
+      "step": 18754
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.144458281444582e-05,
+      "loss": 0.1305,
+      "step": 18755
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.14321295143213e-05,
+      "loss": 0.1238,
+      "step": 18756
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.141967621419675e-05,
+      "loss": 0.1733,
+      "step": 18757
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.140722291407221e-05,
+      "loss": 0.1658,
+      "step": 18758
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.139476961394768e-05,
+      "loss": 0.1162,
+      "step": 18759
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.138231631382316e-05,
+      "loss": 0.1939,
+      "step": 18760
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.136986301369862e-05,
+      "loss": 0.1258,
+      "step": 18761
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.135740971357409e-05,
+      "loss": 0.1291,
+      "step": 18762
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.134495641344955e-05,
+      "loss": 0.198,
+      "step": 18763
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.133250311332502e-05,
+      "loss": 0.2203,
+      "step": 18764
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.13200498132005e-05,
+      "loss": 0.1232,
+      "step": 18765
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.130759651307595e-05,
+      "loss": 0.0925,
+      "step": 18766
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.129514321295143e-05,
+      "loss": 0.0963,
+      "step": 18767
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.12826899128269e-05,
+      "loss": 0.1023,
+      "step": 18768
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.127023661270236e-05,
+      "loss": 0.1761,
+      "step": 18769
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.125778331257783e-05,
+      "loss": 0.236,
+      "step": 18770
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.12453300124533e-05,
+      "loss": 0.1679,
+      "step": 18771
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.123287671232876e-05,
+      "loss": 0.1167,
+      "step": 18772
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.122042341220422e-05,
+      "loss": 0.1231,
+      "step": 18773
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.120797011207968e-05,
+      "loss": 0.146,
+      "step": 18774
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.119551681195515e-05,
+      "loss": 0.1425,
+      "step": 18775
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.118306351183063e-05,
+      "loss": 0.2617,
+      "step": 18776
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.117061021170608e-05,
+      "loss": 0.1405,
+      "step": 18777
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.115815691158156e-05,
+      "loss": 0.1144,
+      "step": 18778
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.114570361145703e-05,
+      "loss": 0.1016,
+      "step": 18779
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.113325031133249e-05,
+      "loss": 0.131,
+      "step": 18780
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.112079701120796e-05,
+      "loss": 0.1498,
+      "step": 18781
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.110834371108343e-05,
+      "loss": 0.1175,
+      "step": 18782
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.10958904109589e-05,
+      "loss": 0.1936,
+      "step": 18783
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.108343711083437e-05,
+      "loss": 0.1276,
+      "step": 18784
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.107098381070984e-05,
+      "loss": 0.135,
+      "step": 18785
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.10585305105853e-05,
+      "loss": 0.1966,
+      "step": 18786
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.104607721046077e-05,
+      "loss": 0.1397,
+      "step": 18787
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.103362391033622e-05,
+      "loss": 0.2445,
+      "step": 18788
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.102117061021169e-05,
+      "loss": 0.1396,
+      "step": 18789
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.100871731008716e-05,
+      "loss": 0.1248,
+      "step": 18790
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.099626400996262e-05,
+      "loss": 0.1669,
+      "step": 18791
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.09838107098381e-05,
+      "loss": 0.2137,
+      "step": 18792
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.097135740971357e-05,
+      "loss": 0.1345,
+      "step": 18793
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.095890410958903e-05,
+      "loss": 0.093,
+      "step": 18794
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.09464508094645e-05,
+      "loss": 0.2399,
+      "step": 18795
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.093399750933997e-05,
+      "loss": 0.1695,
+      "step": 18796
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.092154420921543e-05,
+      "loss": 0.1716,
+      "step": 18797
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.09090909090909e-05,
+      "loss": 0.1233,
+      "step": 18798
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.089663760896638e-05,
+      "loss": 0.0987,
+      "step": 18799
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.088418430884184e-05,
+      "loss": 0.1598,
+      "step": 18800
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.087173100871731e-05,
+      "loss": 0.2408,
+      "step": 18801
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.085927770859278e-05,
+      "loss": 0.1329,
+      "step": 18802
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.084682440846824e-05,
+      "loss": 0.2184,
+      "step": 18803
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.08343711083437e-05,
+      "loss": 0.1865,
+      "step": 18804
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.082191780821916e-05,
+      "loss": 0.1118,
+      "step": 18805
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.080946450809463e-05,
+      "loss": 0.0823,
+      "step": 18806
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.07970112079701e-05,
+      "loss": 0.2249,
+      "step": 18807
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.078455790784556e-05,
+      "loss": 0.1529,
+      "step": 18808
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.077210460772104e-05,
+      "loss": 0.1051,
+      "step": 18809
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.075965130759651e-05,
+      "loss": 0.115,
+      "step": 18810
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.074719800747197e-05,
+      "loss": 0.1538,
+      "step": 18811
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.073474470734744e-05,
+      "loss": 0.2169,
+      "step": 18812
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.072229140722291e-05,
+      "loss": 0.1864,
+      "step": 18813
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.070983810709837e-05,
+      "loss": 0.2355,
+      "step": 18814
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.069738480697384e-05,
+      "loss": 0.1268,
+      "step": 18815
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.068493150684932e-05,
+      "loss": 0.1442,
+      "step": 18816
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.067247820672478e-05,
+      "loss": 0.1702,
+      "step": 18817
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.066002490660025e-05,
+      "loss": 0.0978,
+      "step": 18818
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.064757160647572e-05,
+      "loss": 0.1463,
+      "step": 18819
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.063511830635117e-05,
+      "loss": 0.1607,
+      "step": 18820
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.062266500622664e-05,
+      "loss": 0.1495,
+      "step": 18821
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.06102117061021e-05,
+      "loss": 0.1649,
+      "step": 18822
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.059775840597757e-05,
+      "loss": 0.1678,
+      "step": 18823
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.058530510585305e-05,
+      "loss": 0.1175,
+      "step": 18824
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.05728518057285e-05,
+      "loss": 0.1482,
+      "step": 18825
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.056039850560398e-05,
+      "loss": 0.2215,
+      "step": 18826
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.054794520547945e-05,
+      "loss": 0.1814,
+      "step": 18827
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.053549190535491e-05,
+      "loss": 0.1504,
+      "step": 18828
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.052303860523038e-05,
+      "loss": 0.1131,
+      "step": 18829
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.051058530510585e-05,
+      "loss": 0.155,
+      "step": 18830
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.049813200498131e-05,
+      "loss": 0.1643,
+      "step": 18831
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.048567870485679e-05,
+      "loss": 0.2209,
+      "step": 18832
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.047322540473226e-05,
+      "loss": 0.1594,
+      "step": 18833
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.046077210460772e-05,
+      "loss": 0.1001,
+      "step": 18834
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.044831880448318e-05,
+      "loss": 0.1374,
+      "step": 18835
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.043586550435864e-05,
+      "loss": 0.1779,
+      "step": 18836
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.042341220423411e-05,
+      "loss": 0.1409,
+      "step": 18837
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.041095890410958e-05,
+      "loss": 0.2266,
+      "step": 18838
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.039850560398504e-05,
+      "loss": 0.2059,
+      "step": 18839
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.038605230386051e-05,
+      "loss": 0.1503,
+      "step": 18840
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.037359900373599e-05,
+      "loss": 0.0969,
+      "step": 18841
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.036114570361145e-05,
+      "loss": 0.1912,
+      "step": 18842
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.034869240348692e-05,
+      "loss": 0.147,
+      "step": 18843
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.033623910336238e-05,
+      "loss": 0.0856,
+      "step": 18844
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.032378580323785e-05,
+      "loss": 0.1619,
+      "step": 18845
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.031133250311332e-05,
+      "loss": 0.2109,
+      "step": 18846
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.029887920298878e-05,
+      "loss": 0.1872,
+      "step": 18847
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.028642590286426e-05,
+      "loss": 0.186,
+      "step": 18848
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.027397260273973e-05,
+      "loss": 0.1154,
+      "step": 18849
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.026151930261519e-05,
+      "loss": 0.1258,
+      "step": 18850
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.024906600249065e-05,
+      "loss": 0.3041,
+      "step": 18851
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.023661270236612e-05,
+      "loss": 0.1627,
+      "step": 18852
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.022415940224158e-05,
+      "loss": 0.1191,
+      "step": 18853
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.021170610211705e-05,
+      "loss": 0.1455,
+      "step": 18854
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.019925280199251e-05,
+      "loss": 0.2191,
+      "step": 18855
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.018679950186798e-05,
+      "loss": 0.2425,
+      "step": 18856
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.017434620174346e-05,
+      "loss": 0.1499,
+      "step": 18857
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.016189290161891e-05,
+      "loss": 0.2162,
+      "step": 18858
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.014943960149439e-05,
+      "loss": 0.1783,
+      "step": 18859
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.013698630136986e-05,
+      "loss": 0.1061,
+      "step": 18860
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.012453300124532e-05,
+      "loss": 0.1693,
+      "step": 18861
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.011207970112079e-05,
+      "loss": 0.1346,
+      "step": 18862
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.009962640099626e-05,
+      "loss": 0.2884,
+      "step": 18863
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.008717310087172e-05,
+      "loss": 0.2262,
+      "step": 18864
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.00747198007472e-05,
+      "loss": 0.1908,
+      "step": 18865
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.006226650062267e-05,
+      "loss": 0.1888,
+      "step": 18866
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.004981320049811e-05,
+      "loss": 0.1815,
+      "step": 18867
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.003735990037359e-05,
+      "loss": 0.1644,
+      "step": 18868
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.002490660024905e-05,
+      "loss": 0.0956,
+      "step": 18869
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.001245330012452e-05,
+      "loss": 0.248,
+      "step": 18870
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 0.1815,
+      "step": 18871
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.998754669987545e-05,
+      "loss": 0.129,
+      "step": 18872
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.997509339975092e-05,
+      "loss": 0.1832,
+      "step": 18873
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.99626400996264e-05,
+      "loss": 0.1553,
+      "step": 18874
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.995018679950186e-05,
+      "loss": 0.2149,
+      "step": 18875
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 8.993773349937733e-05,
+      "loss": 0.2415,
+      "step": 18876
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.99252801992528e-05,
+      "loss": 0.2392,
+      "step": 18877
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.991282689912826e-05,
+      "loss": 0.1568,
+      "step": 18878
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.990037359900373e-05,
+      "loss": 0.1193,
+      "step": 18879
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.98879202988792e-05,
+      "loss": 0.1424,
+      "step": 18880
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.987546699875467e-05,
+      "loss": 0.1867,
+      "step": 18881
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.986301369863012e-05,
+      "loss": 0.2291,
+      "step": 18882
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.985056039850558e-05,
+      "loss": 0.1257,
+      "step": 18883
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.983810709838106e-05,
+      "loss": 0.1267,
+      "step": 18884
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.982565379825653e-05,
+      "loss": 0.1846,
+      "step": 18885
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.981320049813199e-05,
+      "loss": 0.1507,
+      "step": 18886
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.980074719800746e-05,
+      "loss": 0.0843,
+      "step": 18887
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.978829389788293e-05,
+      "loss": 0.2312,
+      "step": 18888
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.977584059775839e-05,
+      "loss": 0.1651,
+      "step": 18889
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.976338729763387e-05,
+      "loss": 0.2049,
+      "step": 18890
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.975093399750934e-05,
+      "loss": 0.0732,
+      "step": 18891
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.97384806973848e-05,
+      "loss": 0.1192,
+      "step": 18892
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.972602739726027e-05,
+      "loss": 0.0842,
+      "step": 18893
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.971357409713574e-05,
+      "loss": 0.2347,
+      "step": 18894
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.97011207970112e-05,
+      "loss": 0.1745,
+      "step": 18895
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.968866749688667e-05,
+      "loss": 0.2162,
+      "step": 18896
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.967621419676215e-05,
+      "loss": 0.1956,
+      "step": 18897
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.966376089663759e-05,
+      "loss": 0.1686,
+      "step": 18898
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.965130759651307e-05,
+      "loss": 0.1129,
+      "step": 18899
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.963885429638852e-05,
+      "loss": 0.1009,
+      "step": 18900
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.9626400996264e-05,
+      "loss": 0.2711,
+      "step": 18901
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 8.961394769613947e-05,
+      "loss": 0.2059,
+      "step": 18902
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.960149439601493e-05,
+      "loss": 0.2223,
+      "step": 18903
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.95890410958904e-05,
+      "loss": 0.209,
+      "step": 18904
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.957658779576587e-05,
+      "loss": 0.1258,
+      "step": 18905
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.956413449564133e-05,
+      "loss": 0.188,
+      "step": 18906
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.955168119551681e-05,
+      "loss": 0.1417,
+      "step": 18907
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.953922789539228e-05,
+      "loss": 0.1693,
+      "step": 18908
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.952677459526774e-05,
+      "loss": 0.179,
+      "step": 18909
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.951432129514321e-05,
+      "loss": 0.1343,
+      "step": 18910
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.950186799501868e-05,
+      "loss": 0.1662,
+      "step": 18911
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.948941469489414e-05,
+      "loss": 0.1152,
+      "step": 18912
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.947696139476962e-05,
+      "loss": 0.2971,
+      "step": 18913
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.946450809464506e-05,
+      "loss": 0.1531,
+      "step": 18914
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.945205479452053e-05,
+      "loss": 0.1389,
+      "step": 18915
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.943960149439601e-05,
+      "loss": 0.1961,
+      "step": 18916
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.942714819427147e-05,
+      "loss": 0.1704,
+      "step": 18917
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.941469489414694e-05,
+      "loss": 0.1408,
+      "step": 18918
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.940224159402241e-05,
+      "loss": 0.1135,
+      "step": 18919
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.938978829389787e-05,
+      "loss": 0.1795,
+      "step": 18920
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.937733499377334e-05,
+      "loss": 0.1915,
+      "step": 18921
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.936488169364882e-05,
+      "loss": 0.2047,
+      "step": 18922
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.935242839352428e-05,
+      "loss": 0.1213,
+      "step": 18923
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.933997509339975e-05,
+      "loss": 0.1686,
+      "step": 18924
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.932752179327521e-05,
+      "loss": 0.1024,
+      "step": 18925
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.931506849315068e-05,
+      "loss": 0.1779,
+      "step": 18926
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.930261519302615e-05,
+      "loss": 0.1602,
+      "step": 18927
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 8.929016189290161e-05,
+      "loss": 0.1586,
+      "step": 18928
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.927770859277708e-05,
+      "loss": 0.179,
+      "step": 18929
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.926525529265254e-05,
+      "loss": 0.1502,
+      "step": 18930
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.9252801992528e-05,
+      "loss": 0.1297,
+      "step": 18931
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.924034869240348e-05,
+      "loss": 0.2901,
+      "step": 18932
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.922789539227895e-05,
+      "loss": 0.3348,
+      "step": 18933
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.921544209215441e-05,
+      "loss": 0.1404,
+      "step": 18934
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.920298879202988e-05,
+      "loss": 0.1212,
+      "step": 18935
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.919053549190534e-05,
+      "loss": 0.2203,
+      "step": 18936
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.917808219178081e-05,
+      "loss": 0.2718,
+      "step": 18937
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.916562889165628e-05,
+      "loss": 0.2179,
+      "step": 18938
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.915317559153174e-05,
+      "loss": 0.2344,
+      "step": 18939
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.914072229140722e-05,
+      "loss": 0.1588,
+      "step": 18940
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.912826899128269e-05,
+      "loss": 0.143,
+      "step": 18941
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.911581569115815e-05,
+      "loss": 0.1727,
+      "step": 18942
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.910336239103362e-05,
+      "loss": 0.1126,
+      "step": 18943
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.90909090909091e-05,
+      "loss": 0.112,
+      "step": 18944
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.907845579078454e-05,
+      "loss": 0.2733,
+      "step": 18945
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.906600249066001e-05,
+      "loss": 0.1929,
+      "step": 18946
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.905354919053547e-05,
+      "loss": 0.184,
+      "step": 18947
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.904109589041094e-05,
+      "loss": 0.1394,
+      "step": 18948
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.902864259028642e-05,
+      "loss": 0.0846,
+      "step": 18949
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.901618929016188e-05,
+      "loss": 0.1286,
+      "step": 18950
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.900373599003735e-05,
+      "loss": 0.2621,
+      "step": 18951
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.899128268991282e-05,
+      "loss": 0.1858,
+      "step": 18952
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.897882938978828e-05,
+      "loss": 0.0936,
+      "step": 18953
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 8.896637608966375e-05,
+      "loss": 0.0686,
+      "step": 18954
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.895392278953923e-05,
+      "loss": 0.0952,
+      "step": 18955
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.894146948941469e-05,
+      "loss": 0.1989,
+      "step": 18956
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.892901618929016e-05,
+      "loss": 0.1696,
+      "step": 18957
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.891656288916563e-05,
+      "loss": 0.1958,
+      "step": 18958
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.890410958904109e-05,
+      "loss": 0.17,
+      "step": 18959
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.889165628891656e-05,
+      "loss": 0.1929,
+      "step": 18960
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.887920298879201e-05,
+      "loss": 0.1196,
+      "step": 18961
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.886674968866748e-05,
+      "loss": 0.092,
+      "step": 18962
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.885429638854295e-05,
+      "loss": 0.2248,
+      "step": 18963
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.884184308841841e-05,
+      "loss": 0.1783,
+      "step": 18964
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.882938978829389e-05,
+      "loss": 0.2758,
+      "step": 18965
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.881693648816936e-05,
+      "loss": 0.1303,
+      "step": 18966
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.880448318804482e-05,
+      "loss": 0.1185,
+      "step": 18967
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.879202988792029e-05,
+      "loss": 0.1409,
+      "step": 18968
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.877957658779576e-05,
+      "loss": 0.1811,
+      "step": 18969
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.876712328767122e-05,
+      "loss": 0.19,
+      "step": 18970
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.87546699875467e-05,
+      "loss": 0.1954,
+      "step": 18971
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.874221668742217e-05,
+      "loss": 0.1866,
+      "step": 18972
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.872976338729763e-05,
+      "loss": 0.1535,
+      "step": 18973
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.87173100871731e-05,
+      "loss": 0.1473,
+      "step": 18974
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.870485678704857e-05,
+      "loss": 0.1471,
+      "step": 18975
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.869240348692403e-05,
+      "loss": 0.3605,
+      "step": 18976
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.867995018679949e-05,
+      "loss": 0.184,
+      "step": 18977
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.866749688667495e-05,
+      "loss": 0.2592,
+      "step": 18978
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.865504358655042e-05,
+      "loss": 0.1587,
+      "step": 18979
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 8.86425902864259e-05,
+      "loss": 0.1207,
+      "step": 18980
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.863013698630135e-05,
+      "loss": 0.1701,
+      "step": 18981
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.861768368617683e-05,
+      "loss": 0.1709,
+      "step": 18982
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.86052303860523e-05,
+      "loss": 0.1825,
+      "step": 18983
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.859277708592776e-05,
+      "loss": 0.1405,
+      "step": 18984
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.858032378580323e-05,
+      "loss": 0.2566,
+      "step": 18985
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.85678704856787e-05,
+      "loss": 0.1675,
+      "step": 18986
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.855541718555416e-05,
+      "loss": 0.1067,
+      "step": 18987
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.854296388542964e-05,
+      "loss": 0.141,
+      "step": 18988
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.853051058530511e-05,
+      "loss": 0.1813,
+      "step": 18989
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.851805728518057e-05,
+      "loss": 0.13,
+      "step": 18990
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.850560398505604e-05,
+      "loss": 0.1345,
+      "step": 18991
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.849315068493149e-05,
+      "loss": 0.1351,
+      "step": 18992
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.848069738480696e-05,
+      "loss": 0.1149,
+      "step": 18993
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.846824408468243e-05,
+      "loss": 0.1779,
+      "step": 18994
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.845579078455789e-05,
+      "loss": 0.2004,
+      "step": 18995
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.844333748443336e-05,
+      "loss": 0.1702,
+      "step": 18996
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.843088418430884e-05,
+      "loss": 0.112,
+      "step": 18997
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.84184308841843e-05,
+      "loss": 0.147,
+      "step": 18998
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.840597758405977e-05,
+      "loss": 0.133,
+      "step": 18999
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.839352428393524e-05,
+      "loss": 0.1942,
+      "step": 19000
+    },
+    {
+      "epoch": 7.28,
+      "eval_cer": 0.10177341631386932,
+      "eval_loss": 0.41195443272590637,
+      "eval_runtime": 48.064,
+      "eval_samples_per_second": 36.472,
+      "eval_steps_per_second": 4.577,
+      "eval_wer": 0.16160564782549666,
+      "step": 19000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.83810709838107e-05,
+      "loss": 0.1712,
+      "step": 19001
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.836861768368617e-05,
+      "loss": 0.1895,
+      "step": 19002
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.835616438356165e-05,
+      "loss": 0.1166,
+      "step": 19003
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.83437110834371e-05,
+      "loss": 0.1159,
+      "step": 19004
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.833125778331258e-05,
+      "loss": 0.111,
+      "step": 19005
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 8.831880448318804e-05,
+      "loss": 0.162,
+      "step": 19006
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.830635118306351e-05,
+      "loss": 0.177,
+      "step": 19007
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.829389788293897e-05,
+      "loss": 0.3008,
+      "step": 19008
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.828144458281443e-05,
+      "loss": 0.1708,
+      "step": 19009
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.82689912826899e-05,
+      "loss": 0.0858,
+      "step": 19010
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.825653798256537e-05,
+      "loss": 0.1331,
+      "step": 19011
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.824408468244083e-05,
+      "loss": 0.2151,
+      "step": 19012
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.82316313823163e-05,
+      "loss": 0.2939,
+      "step": 19013
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.821917808219178e-05,
+      "loss": 0.1854,
+      "step": 19014
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.820672478206724e-05,
+      "loss": 0.2207,
+      "step": 19015
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.819427148194271e-05,
+      "loss": 0.1095,
+      "step": 19016
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.818181818181817e-05,
+      "loss": 0.0935,
+      "step": 19017
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.816936488169364e-05,
+      "loss": 0.1274,
+      "step": 19018
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.815691158156911e-05,
+      "loss": 0.1188,
+      "step": 19019
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.814445828144457e-05,
+      "loss": 0.2045,
+      "step": 19020
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.813200498132005e-05,
+      "loss": 0.1203,
+      "step": 19021
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.811955168119552e-05,
+      "loss": 0.1283,
+      "step": 19022
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.810709838107098e-05,
+      "loss": 0.1296,
+      "step": 19023
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.809464508094644e-05,
+      "loss": 0.1038,
+      "step": 19024
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.808219178082191e-05,
+      "loss": 0.1104,
+      "step": 19025
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.806973848069737e-05,
+      "loss": 0.2479,
+      "step": 19026
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.805728518057284e-05,
+      "loss": 0.1649,
+      "step": 19027
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.80448318804483e-05,
+      "loss": 0.1386,
+      "step": 19028
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.803237858032377e-05,
+      "loss": 0.163,
+      "step": 19029
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.801992528019925e-05,
+      "loss": 0.1161,
+      "step": 19030
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.80074719800747e-05,
+      "loss": 0.1356,
+      "step": 19031
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.799501867995018e-05,
+      "loss": 0.1678,
+      "step": 19032
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.798256537982565e-05,
+      "loss": 0.2013,
+      "step": 19033
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.797011207970111e-05,
+      "loss": 0.1426,
+      "step": 19034
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.795765877957658e-05,
+      "loss": 0.1326,
+      "step": 19035
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.794520547945206e-05,
+      "loss": 0.117,
+      "step": 19036
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.793275217932752e-05,
+      "loss": 0.14,
+      "step": 19037
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.792029887920299e-05,
+      "loss": 0.1815,
+      "step": 19038
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.790784557907846e-05,
+      "loss": 0.1907,
+      "step": 19039
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.78953922789539e-05,
+      "loss": 0.1814,
+      "step": 19040
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.788293897882938e-05,
+      "loss": 0.1478,
+      "step": 19041
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.787048567870484e-05,
+      "loss": 0.123,
+      "step": 19042
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.785803237858031e-05,
+      "loss": 0.1187,
+      "step": 19043
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.784557907845578e-05,
+      "loss": 0.1122,
+      "step": 19044
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.783312577833124e-05,
+      "loss": 0.2054,
+      "step": 19045
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.782067247820672e-05,
+      "loss": 0.1585,
+      "step": 19046
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.780821917808219e-05,
+      "loss": 0.273,
+      "step": 19047
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.779576587795765e-05,
+      "loss": 0.209,
+      "step": 19048
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.778331257783312e-05,
+      "loss": 0.1232,
+      "step": 19049
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.777085927770859e-05,
+      "loss": 0.1199,
+      "step": 19050
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.775840597758405e-05,
+      "loss": 0.2133,
+      "step": 19051
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.774595267745952e-05,
+      "loss": 0.1273,
+      "step": 19052
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.7733499377335e-05,
+      "loss": 0.1365,
+      "step": 19053
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.772104607721046e-05,
+      "loss": 0.1009,
+      "step": 19054
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.770859277708592e-05,
+      "loss": 0.1137,
+      "step": 19055
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.769613947696137e-05,
+      "loss": 0.1364,
+      "step": 19056
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.768368617683685e-05,
+      "loss": 0.2197,
+      "step": 19057
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.767123287671232e-05,
+      "loss": 0.1939,
+      "step": 19058
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.765877957658778e-05,
+      "loss": 0.1684,
+      "step": 19059
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.764632627646325e-05,
+      "loss": 0.181,
+      "step": 19060
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.763387297633872e-05,
+      "loss": 0.1087,
+      "step": 19061
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.762141967621418e-05,
+      "loss": 0.1325,
+      "step": 19062
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.760896637608966e-05,
+      "loss": 0.1625,
+      "step": 19063
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.759651307596513e-05,
+      "loss": 0.1442,
+      "step": 19064
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.758405977584059e-05,
+      "loss": 0.1785,
+      "step": 19065
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.757160647571606e-05,
+      "loss": 0.1551,
+      "step": 19066
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.755915317559153e-05,
+      "loss": 0.0928,
+      "step": 19067
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.7546699875467e-05,
+      "loss": 0.1117,
+      "step": 19068
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.753424657534247e-05,
+      "loss": 0.1482,
+      "step": 19069
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.752179327521794e-05,
+      "loss": 0.2014,
+      "step": 19070
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.750933997509338e-05,
+      "loss": 0.1328,
+      "step": 19071
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.749688667496886e-05,
+      "loss": 0.1782,
+      "step": 19072
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.748443337484432e-05,
+      "loss": 0.094,
+      "step": 19073
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.747198007471979e-05,
+      "loss": 0.1217,
+      "step": 19074
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.745952677459526e-05,
+      "loss": 0.2007,
+      "step": 19075
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.744707347447072e-05,
+      "loss": 0.2503,
+      "step": 19076
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.74346201743462e-05,
+      "loss": 0.1644,
+      "step": 19077
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.742216687422167e-05,
+      "loss": 0.2207,
+      "step": 19078
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.740971357409713e-05,
+      "loss": 0.1274,
+      "step": 19079
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.73972602739726e-05,
+      "loss": 0.1607,
+      "step": 19080
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.738480697384807e-05,
+      "loss": 0.1287,
+      "step": 19081
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.737235367372353e-05,
+      "loss": 0.2235,
+      "step": 19082
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.7359900373599e-05,
+      "loss": 0.1721,
+      "step": 19083
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.734744707347448e-05,
+      "loss": 0.1716,
+      "step": 19084
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.733499377334993e-05,
+      "loss": 0.2269,
+      "step": 19085
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.732254047322541e-05,
+      "loss": 0.1528,
+      "step": 19086
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.731008717310085e-05,
+      "loss": 0.1391,
+      "step": 19087
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.729763387297633e-05,
+      "loss": 0.2006,
+      "step": 19088
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.72851805728518e-05,
+      "loss": 0.2628,
+      "step": 19089
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.727272727272726e-05,
+      "loss": 0.146,
+      "step": 19090
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.726027397260273e-05,
+      "loss": 0.1266,
+      "step": 19091
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.72478206724782e-05,
+      "loss": 0.1203,
+      "step": 19092
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.723536737235366e-05,
+      "loss": 0.123,
+      "step": 19093
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.722291407222914e-05,
+      "loss": 0.1645,
+      "step": 19094
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.721046077210461e-05,
+      "loss": 0.3626,
+      "step": 19095
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.719800747198007e-05,
+      "loss": 0.1693,
+      "step": 19096
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.718555417185554e-05,
+      "loss": 0.1096,
+      "step": 19097
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.7173100871731e-05,
+      "loss": 0.0952,
+      "step": 19098
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.716064757160647e-05,
+      "loss": 0.1053,
+      "step": 19099
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.714819427148194e-05,
+      "loss": 0.1466,
+      "step": 19100
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.71357409713574e-05,
+      "loss": 0.1748,
+      "step": 19101
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.712328767123286e-05,
+      "loss": 0.1219,
+      "step": 19102
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.711083437110834e-05,
+      "loss": 0.1614,
+      "step": 19103
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.70983810709838e-05,
+      "loss": 0.1788,
+      "step": 19104
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.708592777085927e-05,
+      "loss": 0.2125,
+      "step": 19105
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.707347447073474e-05,
+      "loss": 0.1662,
+      "step": 19106
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.70610211706102e-05,
+      "loss": 0.2499,
+      "step": 19107
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.704856787048567e-05,
+      "loss": 0.1497,
+      "step": 19108
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.703611457036113e-05,
+      "loss": 0.1508,
+      "step": 19109
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.70236612702366e-05,
+      "loss": 0.1239,
+      "step": 19110
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.701120797011208e-05,
+      "loss": 0.1365,
+      "step": 19111
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.699875466998754e-05,
+      "loss": 0.1932,
+      "step": 19112
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.698630136986301e-05,
+      "loss": 0.3094,
+      "step": 19113
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.697384806973848e-05,
+      "loss": 0.2306,
+      "step": 19114
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.696139476961394e-05,
+      "loss": 0.1683,
+      "step": 19115
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.694894146948941e-05,
+      "loss": 0.2441,
+      "step": 19116
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.693648816936489e-05,
+      "loss": 0.105,
+      "step": 19117
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.692403486924033e-05,
+      "loss": 0.0909,
+      "step": 19118
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.69115815691158e-05,
+      "loss": 0.1018,
+      "step": 19119
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.689912826899126e-05,
+      "loss": 0.1831,
+      "step": 19120
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.688667496886674e-05,
+      "loss": 0.1632,
+      "step": 19121
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.687422166874221e-05,
+      "loss": 0.1468,
+      "step": 19122
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.686176836861767e-05,
+      "loss": 0.1316,
+      "step": 19123
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.684931506849314e-05,
+      "loss": 0.0879,
+      "step": 19124
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.683686176836861e-05,
+      "loss": 0.1066,
+      "step": 19125
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.682440846824407e-05,
+      "loss": 0.3453,
+      "step": 19126
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.681195516811955e-05,
+      "loss": 0.1635,
+      "step": 19127
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.679950186799502e-05,
+      "loss": 0.1463,
+      "step": 19128
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.678704856787048e-05,
+      "loss": 0.1127,
+      "step": 19129
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.677459526774595e-05,
+      "loss": 0.1066,
+      "step": 19130
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.676214196762142e-05,
+      "loss": 0.1863,
+      "step": 19131
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.674968866749688e-05,
+      "loss": 0.2288,
+      "step": 19132
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.673723536737235e-05,
+      "loss": 0.1761,
+      "step": 19133
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.67247820672478e-05,
+      "loss": 0.1604,
+      "step": 19134
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.671232876712327e-05,
+      "loss": 0.1035,
+      "step": 19135
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.669987546699875e-05,
+      "loss": 0.1843,
+      "step": 19136
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.66874221668742e-05,
+      "loss": 0.1416,
+      "step": 19137
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.667496886674968e-05,
+      "loss": 0.2122,
+      "step": 19138
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.666251556662515e-05,
+      "loss": 0.1816,
+      "step": 19139
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.665006226650061e-05,
+      "loss": 0.1903,
+      "step": 19140
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.663760896637608e-05,
+      "loss": 0.2076,
+      "step": 19141
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.662515566625155e-05,
+      "loss": 0.1498,
+      "step": 19142
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.661270236612701e-05,
+      "loss": 0.1864,
+      "step": 19143
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.660024906600249e-05,
+      "loss": 0.0961,
+      "step": 19144
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.658779576587796e-05,
+      "loss": 0.2889,
+      "step": 19145
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.657534246575342e-05,
+      "loss": 0.1832,
+      "step": 19146
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.656288916562889e-05,
+      "loss": 0.1095,
+      "step": 19147
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.655043586550436e-05,
+      "loss": 0.1348,
+      "step": 19148
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.653798256537981e-05,
+      "loss": 0.151,
+      "step": 19149
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.652552926525528e-05,
+      "loss": 0.2048,
+      "step": 19150
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.651307596513074e-05,
+      "loss": 0.2043,
+      "step": 19151
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.650062266500621e-05,
+      "loss": 0.2174,
+      "step": 19152
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.648816936488169e-05,
+      "loss": 0.1286,
+      "step": 19153
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.647571606475715e-05,
+      "loss": 0.1775,
+      "step": 19154
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.646326276463262e-05,
+      "loss": 0.1776,
+      "step": 19155
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.645080946450809e-05,
+      "loss": 0.1703,
+      "step": 19156
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.643835616438355e-05,
+      "loss": 0.2903,
+      "step": 19157
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.642590286425902e-05,
+      "loss": 0.1586,
+      "step": 19158
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.64134495641345e-05,
+      "loss": 0.1155,
+      "step": 19159
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.640099626400996e-05,
+      "loss": 0.1577,
+      "step": 19160
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.638854296388543e-05,
+      "loss": 0.1182,
+      "step": 19161
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.63760896637609e-05,
+      "loss": 0.0928,
+      "step": 19162
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.636363636363636e-05,
+      "loss": 0.3088,
+      "step": 19163
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.635118306351183e-05,
+      "loss": 0.1603,
+      "step": 19164
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.633872976338728e-05,
+      "loss": 0.1631,
+      "step": 19165
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.632627646326275e-05,
+      "loss": 0.0949,
+      "step": 19166
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.631382316313822e-05,
+      "loss": 0.1247,
+      "step": 19167
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.630136986301368e-05,
+      "loss": 0.1016,
+      "step": 19168
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.628891656288916e-05,
+      "loss": 0.188,
+      "step": 19169
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.627646326276463e-05,
+      "loss": 0.2144,
+      "step": 19170
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.626400996264009e-05,
+      "loss": 0.252,
+      "step": 19171
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.625155666251556e-05,
+      "loss": 0.1242,
+      "step": 19172
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.623910336239103e-05,
+      "loss": 0.2002,
+      "step": 19173
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.622665006226649e-05,
+      "loss": 0.1344,
+      "step": 19174
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.621419676214196e-05,
+      "loss": 0.197,
+      "step": 19175
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.620174346201744e-05,
+      "loss": 0.1675,
+      "step": 19176
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.61892901618929e-05,
+      "loss": 0.1222,
+      "step": 19177
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.617683686176837e-05,
+      "loss": 0.185,
+      "step": 19178
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.616438356164383e-05,
+      "loss": 0.1215,
+      "step": 19179
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.61519302615193e-05,
+      "loss": 0.0809,
+      "step": 19180
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.613947696139476e-05,
+      "loss": 0.1351,
+      "step": 19181
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.612702366127022e-05,
+      "loss": 0.2026,
+      "step": 19182
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.611457036114569e-05,
+      "loss": 0.154,
+      "step": 19183
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.610211706102116e-05,
+      "loss": 0.1291,
+      "step": 19184
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.608966376089662e-05,
+      "loss": 0.1536,
+      "step": 19185
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.60772104607721e-05,
+      "loss": 0.1294,
+      "step": 19186
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.606475716064756e-05,
+      "loss": 0.2062,
+      "step": 19187
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.605230386052303e-05,
+      "loss": 0.2299,
+      "step": 19188
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.60398505603985e-05,
+      "loss": 0.1632,
+      "step": 19189
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.602739726027396e-05,
+      "loss": 0.146,
+      "step": 19190
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.601494396014943e-05,
+      "loss": 0.1396,
+      "step": 19191
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.60024906600249e-05,
+      "loss": 0.1248,
+      "step": 19192
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.599003735990037e-05,
+      "loss": 0.1222,
+      "step": 19193
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.597758405977584e-05,
+      "loss": 0.1839,
+      "step": 19194
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.596513075965131e-05,
+      "loss": 0.2278,
+      "step": 19195
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.595267745952677e-05,
+      "loss": 0.125,
+      "step": 19196
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.594022415940223e-05,
+      "loss": 0.1185,
+      "step": 19197
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.592777085927769e-05,
+      "loss": 0.1156,
+      "step": 19198
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.591531755915316e-05,
+      "loss": 0.1271,
+      "step": 19199
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.590286425902863e-05,
+      "loss": 0.0861,
+      "step": 19200
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.589041095890409e-05,
+      "loss": 0.2545,
+      "step": 19201
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.587795765877957e-05,
+      "loss": 0.1404,
+      "step": 19202
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.586550435865504e-05,
+      "loss": 0.1117,
+      "step": 19203
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.58530510585305e-05,
+      "loss": 0.2154,
+      "step": 19204
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.584059775840597e-05,
+      "loss": 0.1311,
+      "step": 19205
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.582814445828144e-05,
+      "loss": 0.2307,
+      "step": 19206
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.58156911581569e-05,
+      "loss": 0.2709,
+      "step": 19207
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.580323785803237e-05,
+      "loss": 0.193,
+      "step": 19208
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.579078455790785e-05,
+      "loss": 0.1722,
+      "step": 19209
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.57783312577833e-05,
+      "loss": 0.1524,
+      "step": 19210
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.576587795765878e-05,
+      "loss": 0.109,
+      "step": 19211
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.575342465753422e-05,
+      "loss": 0.1436,
+      "step": 19212
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.57409713574097e-05,
+      "loss": 0.2476,
+      "step": 19213
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.572851805728517e-05,
+      "loss": 0.2405,
+      "step": 19214
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.571606475716063e-05,
+      "loss": 0.1648,
+      "step": 19215
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.57036114570361e-05,
+      "loss": 0.106,
+      "step": 19216
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.569115815691158e-05,
+      "loss": 0.1448,
+      "step": 19217
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.567870485678703e-05,
+      "loss": 0.1204,
+      "step": 19218
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.566625155666251e-05,
+      "loss": 0.1723,
+      "step": 19219
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.565379825653798e-05,
+      "loss": 0.2034,
+      "step": 19220
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.564134495641344e-05,
+      "loss": 0.1376,
+      "step": 19221
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.562889165628891e-05,
+      "loss": 0.1182,
+      "step": 19222
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.561643835616438e-05,
+      "loss": 0.1561,
+      "step": 19223
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.560398505603984e-05,
+      "loss": 0.1808,
+      "step": 19224
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.559153175591532e-05,
+      "loss": 0.1673,
+      "step": 19225
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.557907845579079e-05,
+      "loss": 0.1578,
+      "step": 19226
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.556662515566625e-05,
+      "loss": 0.2332,
+      "step": 19227
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.555417185554171e-05,
+      "loss": 0.2501,
+      "step": 19228
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.554171855541717e-05,
+      "loss": 0.1394,
+      "step": 19229
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.552926525529264e-05,
+      "loss": 0.1474,
+      "step": 19230
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.551681195516811e-05,
+      "loss": 0.1605,
+      "step": 19231
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.550435865504357e-05,
+      "loss": 0.1653,
+      "step": 19232
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.549190535491904e-05,
+      "loss": 0.1715,
+      "step": 19233
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.547945205479452e-05,
+      "loss": 0.1001,
+      "step": 19234
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.546699875466998e-05,
+      "loss": 0.1225,
+      "step": 19235
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.545454545454545e-05,
+      "loss": 0.156,
+      "step": 19236
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.544209215442092e-05,
+      "loss": 0.1615,
+      "step": 19237
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.542963885429638e-05,
+      "loss": 0.3027,
+      "step": 19238
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.541718555417185e-05,
+      "loss": 0.1663,
+      "step": 19239
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.540473225404733e-05,
+      "loss": 0.1176,
+      "step": 19240
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 8.539227895392278e-05,
+      "loss": 0.1241,
+      "step": 19241
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.537982565379826e-05,
+      "loss": 0.1064,
+      "step": 19242
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.536737235367373e-05,
+      "loss": 0.1987,
+      "step": 19243
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.535491905354918e-05,
+      "loss": 0.182,
+      "step": 19244
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.534246575342465e-05,
+      "loss": 0.2124,
+      "step": 19245
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.533001245330011e-05,
+      "loss": 0.1743,
+      "step": 19246
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.531755915317558e-05,
+      "loss": 0.1366,
+      "step": 19247
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.530510585305105e-05,
+      "loss": 0.1752,
+      "step": 19248
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.529265255292651e-05,
+      "loss": 0.1292,
+      "step": 19249
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.528019925280199e-05,
+      "loss": 0.1266,
+      "step": 19250
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.526774595267746e-05,
+      "loss": 0.1812,
+      "step": 19251
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.525529265255292e-05,
+      "loss": 0.1519,
+      "step": 19252
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.524283935242839e-05,
+      "loss": 0.2965,
+      "step": 19253
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.523038605230386e-05,
+      "loss": 0.1451,
+      "step": 19254
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.521793275217932e-05,
+      "loss": 0.2009,
+      "step": 19255
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.52054794520548e-05,
+      "loss": 0.0777,
+      "step": 19256
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.519302615193027e-05,
+      "loss": 0.3262,
+      "step": 19257
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.518057285180573e-05,
+      "loss": 0.1491,
+      "step": 19258
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.516811955168119e-05,
+      "loss": 0.1283,
+      "step": 19259
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.515566625155664e-05,
+      "loss": 0.104,
+      "step": 19260
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.514321295143212e-05,
+      "loss": 0.2173,
+      "step": 19261
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.513075965130759e-05,
+      "loss": 0.0963,
+      "step": 19262
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.511830635118305e-05,
+      "loss": 0.166,
+      "step": 19263
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.510585305105852e-05,
+      "loss": 0.1764,
+      "step": 19264
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.5093399750934e-05,
+      "loss": 0.1904,
+      "step": 19265
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.508094645080945e-05,
+      "loss": 0.1629,
+      "step": 19266
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 8.506849315068493e-05,
+      "loss": 0.0954,
+      "step": 19267
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.505603985056039e-05,
+      "loss": 0.1229,
+      "step": 19268
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.504358655043586e-05,
+      "loss": 0.2304,
+      "step": 19269
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.503113325031133e-05,
+      "loss": 0.2774,
+      "step": 19270
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.501867995018679e-05,
+      "loss": 0.1841,
+      "step": 19271
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.500622665006226e-05,
+      "loss": 0.1233,
+      "step": 19272
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.499377334993774e-05,
+      "loss": 0.1959,
+      "step": 19273
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.49813200498132e-05,
+      "loss": 0.0967,
+      "step": 19274
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.496886674968865e-05,
+      "loss": 0.0933,
+      "step": 19275
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.495641344956413e-05,
+      "loss": 0.2053,
+      "step": 19276
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.494396014943959e-05,
+      "loss": 0.2038,
+      "step": 19277
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.493150684931506e-05,
+      "loss": 0.1652,
+      "step": 19278
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.491905354919052e-05,
+      "loss": 0.1315,
+      "step": 19279
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.490660024906599e-05,
+      "loss": 0.086,
+      "step": 19280
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.489414694894146e-05,
+      "loss": 0.1608,
+      "step": 19281
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.488169364881692e-05,
+      "loss": 0.1399,
+      "step": 19282
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.48692403486924e-05,
+      "loss": 0.1375,
+      "step": 19283
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.485678704856787e-05,
+      "loss": 0.1681,
+      "step": 19284
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.484433374844333e-05,
+      "loss": 0.205,
+      "step": 19285
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.48318804483188e-05,
+      "loss": 0.122,
+      "step": 19286
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.481942714819427e-05,
+      "loss": 0.1296,
+      "step": 19287
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.480697384806973e-05,
+      "loss": 0.1435,
+      "step": 19288
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.47945205479452e-05,
+      "loss": 0.2213,
+      "step": 19289
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.478206724782068e-05,
+      "loss": 0.1503,
+      "step": 19290
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.476961394769612e-05,
+      "loss": 0.1416,
+      "step": 19291
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.47571606475716e-05,
+      "loss": 0.0877,
+      "step": 19292
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 8.474470734744705e-05,
+      "loss": 0.1737,
+      "step": 19293
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.473225404732253e-05,
+      "loss": 0.153,
+      "step": 19294
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.4719800747198e-05,
+      "loss": 0.2907,
+      "step": 19295
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.470734744707346e-05,
+      "loss": 0.1479,
+      "step": 19296
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.469489414694893e-05,
+      "loss": 0.1374,
+      "step": 19297
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.46824408468244e-05,
+      "loss": 0.1328,
+      "step": 19298
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.466998754669986e-05,
+      "loss": 0.1802,
+      "step": 19299
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.465753424657534e-05,
+      "loss": 0.1653,
+      "step": 19300
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.464508094645081e-05,
+      "loss": 0.2534,
+      "step": 19301
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.463262764632627e-05,
+      "loss": 0.2333,
+      "step": 19302
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.462017434620174e-05,
+      "loss": 0.1859,
+      "step": 19303
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.460772104607721e-05,
+      "loss": 0.2027,
+      "step": 19304
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.459526774595267e-05,
+      "loss": 0.0909,
+      "step": 19305
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.458281444582815e-05,
+      "loss": 0.198,
+      "step": 19306
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.457036114570359e-05,
+      "loss": 0.1444,
+      "step": 19307
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.455790784557906e-05,
+      "loss": 0.1555,
+      "step": 19308
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.454545454545454e-05,
+      "loss": 0.1552,
+      "step": 19309
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.453300124533e-05,
+      "loss": 0.1685,
+      "step": 19310
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.452054794520547e-05,
+      "loss": 0.1811,
+      "step": 19311
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.450809464508094e-05,
+      "loss": 0.1825,
+      "step": 19312
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.44956413449564e-05,
+      "loss": 0.2978,
+      "step": 19313
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.448318804483187e-05,
+      "loss": 0.1713,
+      "step": 19314
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.447073474470735e-05,
+      "loss": 0.1233,
+      "step": 19315
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.44582814445828e-05,
+      "loss": 0.1656,
+      "step": 19316
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.444582814445828e-05,
+      "loss": 0.1405,
+      "step": 19317
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.443337484433375e-05,
+      "loss": 0.1483,
+      "step": 19318
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 8.442092154420921e-05,
+      "loss": 0.1379,
+      "step": 19319
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.440846824408468e-05,
+      "loss": 0.1966,
+      "step": 19320
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.439601494396016e-05,
+      "loss": 0.202,
+      "step": 19321
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.43835616438356e-05,
+      "loss": 0.2038,
+      "step": 19322
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.437110834371107e-05,
+      "loss": 0.1835,
+      "step": 19323
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.435865504358653e-05,
+      "loss": 0.1224,
+      "step": 19324
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.4346201743462e-05,
+      "loss": 0.1595,
+      "step": 19325
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.433374844333748e-05,
+      "loss": 0.1678,
+      "step": 19326
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.432129514321294e-05,
+      "loss": 0.1991,
+      "step": 19327
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.430884184308841e-05,
+      "loss": 0.1615,
+      "step": 19328
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.429638854296388e-05,
+      "loss": 0.1188,
+      "step": 19329
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.428393524283934e-05,
+      "loss": 0.0931,
+      "step": 19330
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.427148194271481e-05,
+      "loss": 0.1311,
+      "step": 19331
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.425902864259029e-05,
+      "loss": 0.3251,
+      "step": 19332
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.424657534246575e-05,
+      "loss": 0.1992,
+      "step": 19333
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.423412204234122e-05,
+      "loss": 0.238,
+      "step": 19334
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.422166874221669e-05,
+      "loss": 0.1067,
+      "step": 19335
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.420921544209215e-05,
+      "loss": 0.1228,
+      "step": 19336
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.419676214196762e-05,
+      "loss": 0.1065,
+      "step": 19337
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.418430884184307e-05,
+      "loss": 0.2341,
+      "step": 19338
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.417185554171854e-05,
+      "loss": 0.2832,
+      "step": 19339
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.415940224159401e-05,
+      "loss": 0.2064,
+      "step": 19340
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.414694894146947e-05,
+      "loss": 0.1915,
+      "step": 19341
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.413449564134495e-05,
+      "loss": 0.1146,
+      "step": 19342
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.412204234122042e-05,
+      "loss": 0.1313,
+      "step": 19343
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.410958904109588e-05,
+      "loss": 0.0901,
+      "step": 19344
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 8.409713574097135e-05,
+      "loss": 0.19,
+      "step": 19345
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.408468244084682e-05,
+      "loss": 0.1551,
+      "step": 19346
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.407222914072228e-05,
+      "loss": 0.1171,
+      "step": 19347
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.405977584059776e-05,
+      "loss": 0.0859,
+      "step": 19348
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.404732254047322e-05,
+      "loss": 0.1004,
+      "step": 19349
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.403486924034869e-05,
+      "loss": 0.1598,
+      "step": 19350
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.402241594022416e-05,
+      "loss": 0.22,
+      "step": 19351
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.400996264009962e-05,
+      "loss": 0.2229,
+      "step": 19352
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.399750933997509e-05,
+      "loss": 0.148,
+      "step": 19353
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.398505603985055e-05,
+      "loss": 0.1694,
+      "step": 19354
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.397260273972601e-05,
+      "loss": 0.124,
+      "step": 19355
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.396014943960148e-05,
+      "loss": 0.1176,
+      "step": 19356
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.394769613947696e-05,
+      "loss": 0.1809,
+      "step": 19357
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.393524283935242e-05,
+      "loss": 0.3063,
+      "step": 19358
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.392278953922789e-05,
+      "loss": 0.1197,
+      "step": 19359
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.391033623910335e-05,
+      "loss": 0.1407,
+      "step": 19360
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.389788293897882e-05,
+      "loss": 0.1247,
+      "step": 19361
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.388542963885429e-05,
+      "loss": 0.0932,
+      "step": 19362
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.387297633872975e-05,
+      "loss": 0.1476,
+      "step": 19363
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.386052303860522e-05,
+      "loss": 0.2099,
+      "step": 19364
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.38480697384807e-05,
+      "loss": 0.1133,
+      "step": 19365
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.383561643835616e-05,
+      "loss": 0.1505,
+      "step": 19366
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.382316313823163e-05,
+      "loss": 0.2002,
+      "step": 19367
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.38107098381071e-05,
+      "loss": 0.0842,
+      "step": 19368
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.379825653798255e-05,
+      "loss": 0.1186,
+      "step": 19369
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.378580323785802e-05,
+      "loss": 0.2114,
+      "step": 19370
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 8.377334993773348e-05,
+      "loss": 0.217,
+      "step": 19371
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.376089663760895e-05,
+      "loss": 0.1598,
+      "step": 19372
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.374844333748443e-05,
+      "loss": 0.1335,
+      "step": 19373
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.373599003735988e-05,
+      "loss": 0.2015,
+      "step": 19374
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.372353673723536e-05,
+      "loss": 0.1099,
+      "step": 19375
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.371108343711083e-05,
+      "loss": 0.2224,
+      "step": 19376
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.369863013698629e-05,
+      "loss": 0.1437,
+      "step": 19377
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.368617683686176e-05,
+      "loss": 0.1879,
+      "step": 19378
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.367372353673723e-05,
+      "loss": 0.195,
+      "step": 19379
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.36612702366127e-05,
+      "loss": 0.1123,
+      "step": 19380
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.364881693648817e-05,
+      "loss": 0.1043,
+      "step": 19381
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.363636363636364e-05,
+      "loss": 0.2093,
+      "step": 19382
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.36239103362391e-05,
+      "loss": 0.1905,
+      "step": 19383
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.361145703611457e-05,
+      "loss": 0.1465,
+      "step": 19384
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.359900373599002e-05,
+      "loss": 0.1753,
+      "step": 19385
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.358655043586549e-05,
+      "loss": 0.0909,
+      "step": 19386
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.357409713574096e-05,
+      "loss": 0.1164,
+      "step": 19387
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.356164383561642e-05,
+      "loss": 0.2423,
+      "step": 19388
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.35491905354919e-05,
+      "loss": 0.1875,
+      "step": 19389
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.353673723536737e-05,
+      "loss": 0.1353,
+      "step": 19390
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.352428393524283e-05,
+      "loss": 0.1197,
+      "step": 19391
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.35118306351183e-05,
+      "loss": 0.1089,
+      "step": 19392
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.349937733499377e-05,
+      "loss": 0.1031,
+      "step": 19393
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.348692403486923e-05,
+      "loss": 0.1432,
+      "step": 19394
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.34744707347447e-05,
+      "loss": 0.334,
+      "step": 19395
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.346201743462018e-05,
+      "loss": 0.1533,
+      "step": 19396
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 8.344956413449563e-05,
+      "loss": 0.2063,
+      "step": 19397
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.343711083437111e-05,
+      "loss": 0.1245,
+      "step": 19398
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.342465753424658e-05,
+      "loss": 0.174,
+      "step": 19399
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.341220423412204e-05,
+      "loss": 0.1306,
+      "step": 19400
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.33997509339975e-05,
+      "loss": 0.2713,
+      "step": 19401
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.338729763387296e-05,
+      "loss": 0.1679,
+      "step": 19402
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.337484433374843e-05,
+      "loss": 0.1527,
+      "step": 19403
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.33623910336239e-05,
+      "loss": 0.1672,
+      "step": 19404
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.334993773349936e-05,
+      "loss": 0.2359,
+      "step": 19405
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.333748443337484e-05,
+      "loss": 0.093,
+      "step": 19406
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.332503113325031e-05,
+      "loss": 0.2089,
+      "step": 19407
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.331257783312577e-05,
+      "loss": 0.1643,
+      "step": 19408
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.330012453300124e-05,
+      "loss": 0.1808,
+      "step": 19409
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.328767123287671e-05,
+      "loss": 0.1415,
+      "step": 19410
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.327521793275217e-05,
+      "loss": 0.0889,
+      "step": 19411
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.326276463262764e-05,
+      "loss": 0.1403,
+      "step": 19412
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.325031133250312e-05,
+      "loss": 0.2923,
+      "step": 19413
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.323785803237858e-05,
+      "loss": 0.1581,
+      "step": 19414
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.322540473225405e-05,
+      "loss": 0.128,
+      "step": 19415
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.32129514321295e-05,
+      "loss": 0.1256,
+      "step": 19416
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.320049813200497e-05,
+      "loss": 0.1828,
+      "step": 19417
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.318804483188044e-05,
+      "loss": 0.1179,
+      "step": 19418
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.31755915317559e-05,
+      "loss": 0.1589,
+      "step": 19419
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.316313823163137e-05,
+      "loss": 0.2422,
+      "step": 19420
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.315068493150684e-05,
+      "loss": 0.1881,
+      "step": 19421
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.31382316313823e-05,
+      "loss": 0.1129,
+      "step": 19422
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.312577833125778e-05,
+      "loss": 0.1486,
+      "step": 19423
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 8.311332503113325e-05,
+      "loss": 0.1153,
+      "step": 19424
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.310087173100871e-05,
+      "loss": 0.2768,
+      "step": 19425
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.308841843088418e-05,
+      "loss": 0.2432,
+      "step": 19426
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.307596513075965e-05,
+      "loss": 0.1855,
+      "step": 19427
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.306351183063511e-05,
+      "loss": 0.1297,
+      "step": 19428
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.305105853051059e-05,
+      "loss": 0.1241,
+      "step": 19429
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.303860523038604e-05,
+      "loss": 0.1517,
+      "step": 19430
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.302615193026152e-05,
+      "loss": 0.1043,
+      "step": 19431
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.301369863013698e-05,
+      "loss": 0.2266,
+      "step": 19432
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.300124533001244e-05,
+      "loss": 0.1498,
+      "step": 19433
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.298879202988791e-05,
+      "loss": 0.1375,
+      "step": 19434
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.297633872976338e-05,
+      "loss": 0.1558,
+      "step": 19435
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.296388542963884e-05,
+      "loss": 0.1514,
+      "step": 19436
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.295143212951431e-05,
+      "loss": 0.1089,
+      "step": 19437
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.293897882938979e-05,
+      "loss": 0.3083,
+      "step": 19438
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.292652552926525e-05,
+      "loss": 0.2339,
+      "step": 19439
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.291407222914072e-05,
+      "loss": 0.1206,
+      "step": 19440
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.290161892901618e-05,
+      "loss": 0.1595,
+      "step": 19441
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.288916562889165e-05,
+      "loss": 0.112,
+      "step": 19442
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.287671232876712e-05,
+      "loss": 0.117,
+      "step": 19443
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.286425902864258e-05,
+      "loss": 0.1621,
+      "step": 19444
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.285180572851805e-05,
+      "loss": 0.1728,
+      "step": 19445
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.283935242839353e-05,
+      "loss": 0.2392,
+      "step": 19446
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.282689912826899e-05,
+      "loss": 0.1401,
+      "step": 19447
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.281444582814445e-05,
+      "loss": 0.1632,
+      "step": 19448
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.280199252801992e-05,
+      "loss": 0.1413,
+      "step": 19449
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 8.278953922789538e-05,
+      "loss": 0.1501,
+      "step": 19450
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.277708592777085e-05,
+      "loss": 0.2216,
+      "step": 19451
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.276463262764631e-05,
+      "loss": 0.2185,
+      "step": 19452
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.275217932752178e-05,
+      "loss": 0.1313,
+      "step": 19453
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.273972602739725e-05,
+      "loss": 0.1145,
+      "step": 19454
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.272727272727271e-05,
+      "loss": 0.1432,
+      "step": 19455
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.271481942714819e-05,
+      "loss": 0.1485,
+      "step": 19456
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.270236612702366e-05,
+      "loss": 0.2542,
+      "step": 19457
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.268991282689912e-05,
+      "loss": 0.1061,
+      "step": 19458
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.267745952677459e-05,
+      "loss": 0.1691,
+      "step": 19459
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.266500622665006e-05,
+      "loss": 0.1536,
+      "step": 19460
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.265255292652552e-05,
+      "loss": 0.1402,
+      "step": 19461
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.2640099626401e-05,
+      "loss": 0.1981,
+      "step": 19462
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.262764632627647e-05,
+      "loss": 0.175,
+      "step": 19463
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.261519302615191e-05,
+      "loss": 0.1824,
+      "step": 19464
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.260273972602739e-05,
+      "loss": 0.0988,
+      "step": 19465
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.259028642590285e-05,
+      "loss": 0.0917,
+      "step": 19466
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.257783312577832e-05,
+      "loss": 0.128,
+      "step": 19467
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.256537982565379e-05,
+      "loss": 0.1478,
+      "step": 19468
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.255292652552925e-05,
+      "loss": 0.1741,
+      "step": 19469
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.254047322540472e-05,
+      "loss": 0.2369,
+      "step": 19470
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.25280199252802e-05,
+      "loss": 0.1174,
+      "step": 19471
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.251556662515566e-05,
+      "loss": 0.1253,
+      "step": 19472
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.250311332503113e-05,
+      "loss": 0.1161,
+      "step": 19473
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.24906600249066e-05,
+      "loss": 0.2484,
+      "step": 19474
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.247820672478206e-05,
+      "loss": 0.0898,
+      "step": 19475
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 8.246575342465753e-05,
+      "loss": 0.3499,
+      "step": 19476
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.2453300124533e-05,
+      "loss": 0.2323,
+      "step": 19477
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.244084682440846e-05,
+      "loss": 0.1312,
+      "step": 19478
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.242839352428392e-05,
+      "loss": 0.172,
+      "step": 19479
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.241594022415938e-05,
+      "loss": 0.092,
+      "step": 19480
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.240348692403486e-05,
+      "loss": 0.2709,
+      "step": 19481
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.239103362391033e-05,
+      "loss": 0.2038,
+      "step": 19482
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.237858032378579e-05,
+      "loss": 0.1405,
+      "step": 19483
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.236612702366126e-05,
+      "loss": 0.1226,
+      "step": 19484
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.235367372353673e-05,
+      "loss": 0.1991,
+      "step": 19485
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.234122042341219e-05,
+      "loss": 0.2076,
+      "step": 19486
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.232876712328766e-05,
+      "loss": 0.2265,
+      "step": 19487
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.231631382316314e-05,
+      "loss": 0.1617,
+      "step": 19488
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.23038605230386e-05,
+      "loss": 0.1482,
+      "step": 19489
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.229140722291407e-05,
+      "loss": 0.1811,
+      "step": 19490
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.227895392278954e-05,
+      "loss": 0.1576,
+      "step": 19491
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.2266500622665e-05,
+      "loss": 0.1896,
+      "step": 19492
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.225404732254047e-05,
+      "loss": 0.1566,
+      "step": 19493
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.224159402241595e-05,
+      "loss": 0.1708,
+      "step": 19494
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.222914072229139e-05,
+      "loss": 0.1459,
+      "step": 19495
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.221668742216687e-05,
+      "loss": 0.1276,
+      "step": 19496
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.220423412204232e-05,
+      "loss": 0.2253,
+      "step": 19497
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.21917808219178e-05,
+      "loss": 0.0967,
+      "step": 19498
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.217932752179327e-05,
+      "loss": 0.1788,
+      "step": 19499
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.216687422166873e-05,
+      "loss": 0.1517,
+      "step": 19500
+    },
+    {
+      "epoch": 7.47,
+      "eval_cer": 0.10199972686657433,
+      "eval_loss": 0.38555046916007996,
+      "eval_runtime": 47.3526,
+      "eval_samples_per_second": 37.02,
+      "eval_steps_per_second": 4.646,
+      "eval_wer": 0.16346110331719238,
+      "step": 19500
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.21544209215442e-05,
+      "loss": 0.1745,
+      "step": 19501
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 8.214196762141967e-05,
+      "loss": 0.2573,
+      "step": 19502
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.212951432129513e-05,
+      "loss": 0.1428,
+      "step": 19503
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.21170610211706e-05,
+      "loss": 0.1464,
+      "step": 19504
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.210460772104608e-05,
+      "loss": 0.1687,
+      "step": 19505
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.209215442092154e-05,
+      "loss": 0.1153,
+      "step": 19506
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.207970112079701e-05,
+      "loss": 0.1923,
+      "step": 19507
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.206724782067248e-05,
+      "loss": 0.166,
+      "step": 19508
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.205479452054794e-05,
+      "loss": 0.1686,
+      "step": 19509
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.204234122042342e-05,
+      "loss": 0.1359,
+      "step": 19510
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.202988792029886e-05,
+      "loss": 0.2578,
+      "step": 19511
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.201743462017433e-05,
+      "loss": 0.0958,
+      "step": 19512
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.20049813200498e-05,
+      "loss": 0.246,
+      "step": 19513
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.199252801992527e-05,
+      "loss": 0.1936,
+      "step": 19514
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.198007471980074e-05,
+      "loss": 0.1698,
+      "step": 19515
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.196762141967621e-05,
+      "loss": 0.1201,
+      "step": 19516
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.195516811955167e-05,
+      "loss": 0.0971,
+      "step": 19517
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.194271481942714e-05,
+      "loss": 0.0824,
+      "step": 19518
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.193026151930262e-05,
+      "loss": 0.1344,
+      "step": 19519
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.191780821917807e-05,
+      "loss": 0.2274,
+      "step": 19520
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.190535491905355e-05,
+      "loss": 0.2359,
+      "step": 19521
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.1892901618929e-05,
+      "loss": 0.1319,
+      "step": 19522
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.188044831880448e-05,
+      "loss": 0.2515,
+      "step": 19523
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.186799501867995e-05,
+      "loss": 0.1265,
+      "step": 19524
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.185554171855541e-05,
+      "loss": 0.1587,
+      "step": 19525
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.184308841843087e-05,
+      "loss": 0.1714,
+      "step": 19526
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.183063511830634e-05,
+      "loss": 0.2047,
+      "step": 19527
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 8.18181818181818e-05,
+      "loss": 0.2074,
+      "step": 19528
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.180572851805728e-05,
+      "loss": 0.1216,
+      "step": 19529
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.179327521793275e-05,
+      "loss": 0.1167,
+      "step": 19530
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.178082191780821e-05,
+      "loss": 0.0755,
+      "step": 19531
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.176836861768368e-05,
+      "loss": 0.227,
+      "step": 19532
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.175591531755914e-05,
+      "loss": 0.1885,
+      "step": 19533
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.174346201743461e-05,
+      "loss": 0.1354,
+      "step": 19534
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.173100871731008e-05,
+      "loss": 0.1154,
+      "step": 19535
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.171855541718554e-05,
+      "loss": 0.1082,
+      "step": 19536
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.170610211706102e-05,
+      "loss": 0.1678,
+      "step": 19537
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.169364881693649e-05,
+      "loss": 0.2283,
+      "step": 19538
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.168119551681195e-05,
+      "loss": 0.1612,
+      "step": 19539
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.166874221668742e-05,
+      "loss": 0.1715,
+      "step": 19540
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.16562889165629e-05,
+      "loss": 0.1999,
+      "step": 19541
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.164383561643834e-05,
+      "loss": 0.1146,
+      "step": 19542
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.163138231631381e-05,
+      "loss": 0.1668,
+      "step": 19543
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.161892901618927e-05,
+      "loss": 0.1259,
+      "step": 19544
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.160647571606474e-05,
+      "loss": 0.2109,
+      "step": 19545
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.159402241594022e-05,
+      "loss": 0.1593,
+      "step": 19546
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.158156911581568e-05,
+      "loss": 0.175,
+      "step": 19547
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.156911581569115e-05,
+      "loss": 0.1298,
+      "step": 19548
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.155666251556662e-05,
+      "loss": 0.1103,
+      "step": 19549
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.154420921544208e-05,
+      "loss": 0.1641,
+      "step": 19550
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.153175591531755e-05,
+      "loss": 0.1983,
+      "step": 19551
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.151930261519303e-05,
+      "loss": 0.1134,
+      "step": 19552
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.150684931506848e-05,
+      "loss": 0.1418,
+      "step": 19553
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 8.149439601494396e-05,
+      "loss": 0.1788,
+      "step": 19554
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.148194271481943e-05,
+      "loss": 0.22,
+      "step": 19555
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.146948941469489e-05,
+      "loss": 0.1236,
+      "step": 19556
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.145703611457036e-05,
+      "loss": 0.2897,
+      "step": 19557
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.144458281444581e-05,
+      "loss": 0.2571,
+      "step": 19558
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.143212951432128e-05,
+      "loss": 0.107,
+      "step": 19559
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.141967621419675e-05,
+      "loss": 0.1476,
+      "step": 19560
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.140722291407221e-05,
+      "loss": 0.0843,
+      "step": 19561
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.139476961394769e-05,
+      "loss": 0.0967,
+      "step": 19562
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.138231631382316e-05,
+      "loss": 0.1345,
+      "step": 19563
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.136986301369862e-05,
+      "loss": 0.1494,
+      "step": 19564
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.135740971357409e-05,
+      "loss": 0.1695,
+      "step": 19565
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.134495641344956e-05,
+      "loss": 0.2116,
+      "step": 19566
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.133250311332502e-05,
+      "loss": 0.1675,
+      "step": 19567
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.13200498132005e-05,
+      "loss": 0.0928,
+      "step": 19568
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.130759651307597e-05,
+      "loss": 0.1779,
+      "step": 19569
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.129514321295143e-05,
+      "loss": 0.2728,
+      "step": 19570
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.12826899128269e-05,
+      "loss": 0.1875,
+      "step": 19571
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.127023661270237e-05,
+      "loss": 0.0938,
+      "step": 19572
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.125778331257782e-05,
+      "loss": 0.1715,
+      "step": 19573
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.124533001245329e-05,
+      "loss": 0.1051,
+      "step": 19574
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.123287671232875e-05,
+      "loss": 0.1346,
+      "step": 19575
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.122042341220422e-05,
+      "loss": 0.1837,
+      "step": 19576
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.12079701120797e-05,
+      "loss": 0.1857,
+      "step": 19577
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.119551681195515e-05,
+      "loss": 0.343,
+      "step": 19578
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.118306351183063e-05,
+      "loss": 0.1244,
+      "step": 19579
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 8.11706102117061e-05,
+      "loss": 0.0962,
+      "step": 19580
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.115815691158156e-05,
+      "loss": 0.1695,
+      "step": 19581
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.114570361145703e-05,
+      "loss": 0.2417,
+      "step": 19582
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.11332503113325e-05,
+      "loss": 0.1989,
+      "step": 19583
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.112079701120796e-05,
+      "loss": 0.1563,
+      "step": 19584
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.110834371108344e-05,
+      "loss": 0.1741,
+      "step": 19585
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.109589041095891e-05,
+      "loss": 0.1368,
+      "step": 19586
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.108343711083437e-05,
+      "loss": 0.1006,
+      "step": 19587
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.107098381070984e-05,
+      "loss": 0.1724,
+      "step": 19588
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.105853051058529e-05,
+      "loss": 0.1984,
+      "step": 19589
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.104607721046076e-05,
+      "loss": 0.1777,
+      "step": 19590
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.103362391033623e-05,
+      "loss": 0.1809,
+      "step": 19591
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.102117061021169e-05,
+      "loss": 0.1826,
+      "step": 19592
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.100871731008716e-05,
+      "loss": 0.0974,
+      "step": 19593
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.099626400996264e-05,
+      "loss": 0.1916,
+      "step": 19594
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.09838107098381e-05,
+      "loss": 0.2098,
+      "step": 19595
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.097135740971357e-05,
+      "loss": 0.1906,
+      "step": 19596
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.095890410958904e-05,
+      "loss": 0.1861,
+      "step": 19597
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.09464508094645e-05,
+      "loss": 0.1408,
+      "step": 19598
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.093399750933997e-05,
+      "loss": 0.1059,
+      "step": 19599
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.092154420921545e-05,
+      "loss": 0.0578,
+      "step": 19600
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.09090909090909e-05,
+      "loss": 0.1753,
+      "step": 19601
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.089663760896638e-05,
+      "loss": 0.1938,
+      "step": 19602
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.088418430884184e-05,
+      "loss": 0.1473,
+      "step": 19603
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.087173100871731e-05,
+      "loss": 0.1816,
+      "step": 19604
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.085927770859277e-05,
+      "loss": 0.1251,
+      "step": 19605
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 8.084682440846823e-05,
+      "loss": 0.1414,
+      "step": 19606
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.08343711083437e-05,
+      "loss": 0.1975,
+      "step": 19607
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.082191780821917e-05,
+      "loss": 0.1608,
+      "step": 19608
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.080946450809463e-05,
+      "loss": 0.1308,
+      "step": 19609
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.07970112079701e-05,
+      "loss": 0.12,
+      "step": 19610
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.078455790784558e-05,
+      "loss": 0.1792,
+      "step": 19611
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.077210460772104e-05,
+      "loss": 0.1525,
+      "step": 19612
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.075965130759651e-05,
+      "loss": 0.1519,
+      "step": 19613
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.074719800747197e-05,
+      "loss": 0.1611,
+      "step": 19614
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.073474470734744e-05,
+      "loss": 0.1735,
+      "step": 19615
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.072229140722291e-05,
+      "loss": 0.1168,
+      "step": 19616
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.070983810709837e-05,
+      "loss": 0.1151,
+      "step": 19617
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.069738480697385e-05,
+      "loss": 0.124,
+      "step": 19618
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.068493150684932e-05,
+      "loss": 0.1378,
+      "step": 19619
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.067247820672478e-05,
+      "loss": 0.2435,
+      "step": 19620
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.066002490660024e-05,
+      "loss": 0.1697,
+      "step": 19621
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.06475716064757e-05,
+      "loss": 0.1098,
+      "step": 19622
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.063511830635117e-05,
+      "loss": 0.1358,
+      "step": 19623
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.062266500622664e-05,
+      "loss": 0.1538,
+      "step": 19624
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.06102117061021e-05,
+      "loss": 0.119,
+      "step": 19625
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.059775840597757e-05,
+      "loss": 0.2734,
+      "step": 19626
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.058530510585305e-05,
+      "loss": 0.1488,
+      "step": 19627
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.05728518057285e-05,
+      "loss": 0.1871,
+      "step": 19628
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.056039850560398e-05,
+      "loss": 0.0798,
+      "step": 19629
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.054794520547945e-05,
+      "loss": 0.1172,
+      "step": 19630
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.053549190535491e-05,
+      "loss": 0.0772,
+      "step": 19631
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 8.052303860523038e-05,
+      "loss": 0.1868,
+      "step": 19632
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.051058530510586e-05,
+      "loss": 0.1746,
+      "step": 19633
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.049813200498131e-05,
+      "loss": 0.2276,
+      "step": 19634
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.048567870485679e-05,
+      "loss": 0.1913,
+      "step": 19635
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.047322540473223e-05,
+      "loss": 0.1225,
+      "step": 19636
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.04607721046077e-05,
+      "loss": 0.1125,
+      "step": 19637
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.044831880448318e-05,
+      "loss": 0.1762,
+      "step": 19638
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.043586550435864e-05,
+      "loss": 0.23,
+      "step": 19639
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.042341220423411e-05,
+      "loss": 0.172,
+      "step": 19640
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.041095890410958e-05,
+      "loss": 0.1619,
+      "step": 19641
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.039850560398504e-05,
+      "loss": 0.1264,
+      "step": 19642
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.038605230386051e-05,
+      "loss": 0.1091,
+      "step": 19643
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.037359900373599e-05,
+      "loss": 0.143,
+      "step": 19644
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.036114570361145e-05,
+      "loss": 0.1952,
+      "step": 19645
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.034869240348692e-05,
+      "loss": 0.1376,
+      "step": 19646
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.033623910336239e-05,
+      "loss": 0.1848,
+      "step": 19647
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.032378580323785e-05,
+      "loss": 0.1609,
+      "step": 19648
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.031133250311332e-05,
+      "loss": 0.0791,
+      "step": 19649
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.02988792029888e-05,
+      "loss": 0.1359,
+      "step": 19650
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.028642590286426e-05,
+      "loss": 0.2839,
+      "step": 19651
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.027397260273972e-05,
+      "loss": 0.1735,
+      "step": 19652
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.026151930261517e-05,
+      "loss": 0.1983,
+      "step": 19653
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.024906600249065e-05,
+      "loss": 0.1521,
+      "step": 19654
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.023661270236612e-05,
+      "loss": 0.1447,
+      "step": 19655
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.022415940224158e-05,
+      "loss": 0.1551,
+      "step": 19656
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.021170610211705e-05,
+      "loss": 0.23,
+      "step": 19657
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 8.019925280199252e-05,
+      "loss": 0.1721,
+      "step": 19658
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.018679950186798e-05,
+      "loss": 0.2111,
+      "step": 19659
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.017434620174346e-05,
+      "loss": 0.1142,
+      "step": 19660
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.016189290161893e-05,
+      "loss": 0.1427,
+      "step": 19661
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.014943960149439e-05,
+      "loss": 0.1963,
+      "step": 19662
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.013698630136986e-05,
+      "loss": 0.1694,
+      "step": 19663
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.012453300124533e-05,
+      "loss": 0.2531,
+      "step": 19664
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.011207970112079e-05,
+      "loss": 0.1663,
+      "step": 19665
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.009962640099627e-05,
+      "loss": 0.1405,
+      "step": 19666
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.008717310087174e-05,
+      "loss": 0.1337,
+      "step": 19667
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.007471980074718e-05,
+      "loss": 0.1318,
+      "step": 19668
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.006226650062266e-05,
+      "loss": 0.1811,
+      "step": 19669
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.004981320049812e-05,
+      "loss": 0.2501,
+      "step": 19670
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.003735990037359e-05,
+      "loss": 0.1466,
+      "step": 19671
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.002490660024906e-05,
+      "loss": 0.1709,
+      "step": 19672
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 8.001245330012452e-05,
+      "loss": 0.151,
+      "step": 19673
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.999999999999999e-05,
+      "loss": 0.2142,
+      "step": 19674
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.998754669987547e-05,
+      "loss": 0.1399,
+      "step": 19675
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.997509339975092e-05,
+      "loss": 0.2058,
+      "step": 19676
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.99626400996264e-05,
+      "loss": 0.1474,
+      "step": 19677
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.995018679950187e-05,
+      "loss": 0.1632,
+      "step": 19678
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.993773349937733e-05,
+      "loss": 0.1333,
+      "step": 19679
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.99252801992528e-05,
+      "loss": 0.2364,
+      "step": 19680
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.991282689912827e-05,
+      "loss": 0.1598,
+      "step": 19681
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.990037359900373e-05,
+      "loss": 0.2966,
+      "step": 19682
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.98879202988792e-05,
+      "loss": 0.2443,
+      "step": 19683
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.987546699875465e-05,
+      "loss": 0.2032,
+      "step": 19684
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.986301369863013e-05,
+      "loss": 0.093,
+      "step": 19685
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.98505603985056e-05,
+      "loss": 0.1004,
+      "step": 19686
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.983810709838106e-05,
+      "loss": 0.1146,
+      "step": 19687
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.982565379825653e-05,
+      "loss": 0.1734,
+      "step": 19688
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.9813200498132e-05,
+      "loss": 0.1792,
+      "step": 19689
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.980074719800746e-05,
+      "loss": 0.2823,
+      "step": 19690
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.978829389788293e-05,
+      "loss": 0.1597,
+      "step": 19691
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.97758405977584e-05,
+      "loss": 0.1073,
+      "step": 19692
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.976338729763387e-05,
+      "loss": 0.0829,
+      "step": 19693
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.975093399750934e-05,
+      "loss": 0.2456,
+      "step": 19694
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.97384806973848e-05,
+      "loss": 0.2741,
+      "step": 19695
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.972602739726027e-05,
+      "loss": 0.1832,
+      "step": 19696
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.971357409713574e-05,
+      "loss": 0.1252,
+      "step": 19697
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.97011207970112e-05,
+      "loss": 0.1385,
+      "step": 19698
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.968866749688666e-05,
+      "loss": 0.1732,
+      "step": 19699
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.967621419676213e-05,
+      "loss": 0.1132,
+      "step": 19700
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.96637608966376e-05,
+      "loss": 0.1879,
+      "step": 19701
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.965130759651307e-05,
+      "loss": 0.1809,
+      "step": 19702
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.963885429638853e-05,
+      "loss": 0.1331,
+      "step": 19703
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.9626400996264e-05,
+      "loss": 0.1447,
+      "step": 19704
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.961394769613947e-05,
+      "loss": 0.1358,
+      "step": 19705
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.960149439601493e-05,
+      "loss": 0.2198,
+      "step": 19706
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.95890410958904e-05,
+      "loss": 0.2073,
+      "step": 19707
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.957658779576588e-05,
+      "loss": 0.1654,
+      "step": 19708
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.956413449564133e-05,
+      "loss": 0.1075,
+      "step": 19709
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 7.955168119551681e-05,
+      "loss": 0.0778,
+      "step": 19710
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.953922789539228e-05,
+      "loss": 0.1298,
+      "step": 19711
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.952677459526774e-05,
+      "loss": 0.2249,
+      "step": 19712
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.951432129514321e-05,
+      "loss": 0.1829,
+      "step": 19713
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.950186799501869e-05,
+      "loss": 0.2153,
+      "step": 19714
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.948941469489413e-05,
+      "loss": 0.258,
+      "step": 19715
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.94769613947696e-05,
+      "loss": 0.2052,
+      "step": 19716
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.946450809464506e-05,
+      "loss": 0.1616,
+      "step": 19717
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.945205479452054e-05,
+      "loss": 0.0648,
+      "step": 19718
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.943960149439601e-05,
+      "loss": 0.1322,
+      "step": 19719
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.942714819427147e-05,
+      "loss": 0.2388,
+      "step": 19720
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.941469489414694e-05,
+      "loss": 0.1802,
+      "step": 19721
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.940224159402241e-05,
+      "loss": 0.1082,
+      "step": 19722
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.938978829389787e-05,
+      "loss": 0.1203,
+      "step": 19723
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.937733499377334e-05,
+      "loss": 0.1455,
+      "step": 19724
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.936488169364882e-05,
+      "loss": 0.1166,
+      "step": 19725
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.935242839352428e-05,
+      "loss": 0.2753,
+      "step": 19726
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.933997509339975e-05,
+      "loss": 0.1295,
+      "step": 19727
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.932752179327522e-05,
+      "loss": 0.1928,
+      "step": 19728
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.931506849315068e-05,
+      "loss": 0.1341,
+      "step": 19729
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.930261519302615e-05,
+      "loss": 0.1091,
+      "step": 19730
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.92901618929016e-05,
+      "loss": 0.1407,
+      "step": 19731
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.927770859277707e-05,
+      "loss": 0.1855,
+      "step": 19732
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.926525529265254e-05,
+      "loss": 0.1773,
+      "step": 19733
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.9252801992528e-05,
+      "loss": 0.1566,
+      "step": 19734
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.924034869240348e-05,
+      "loss": 0.1282,
+      "step": 19735
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.922789539227895e-05,
+      "loss": 0.0821,
+      "step": 19736
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 7.921544209215441e-05,
+      "loss": 0.182,
+      "step": 19737
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.920298879202988e-05,
+      "loss": 0.1909,
+      "step": 19738
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.919053549190535e-05,
+      "loss": 0.1367,
+      "step": 19739
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.917808219178081e-05,
+      "loss": 0.1267,
+      "step": 19740
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.916562889165629e-05,
+      "loss": 0.1778,
+      "step": 19741
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.915317559153176e-05,
+      "loss": 0.1228,
+      "step": 19742
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.914072229140722e-05,
+      "loss": 0.1054,
+      "step": 19743
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.912826899128269e-05,
+      "loss": 0.1125,
+      "step": 19744
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.911581569115816e-05,
+      "loss": 0.2176,
+      "step": 19745
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.910336239103361e-05,
+      "loss": 0.2309,
+      "step": 19746
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.909090909090908e-05,
+      "loss": 0.2363,
+      "step": 19747
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.907845579078454e-05,
+      "loss": 0.1235,
+      "step": 19748
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.906600249066001e-05,
+      "loss": 0.1704,
+      "step": 19749
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.905354919053549e-05,
+      "loss": 0.2303,
+      "step": 19750
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.904109589041095e-05,
+      "loss": 0.2145,
+      "step": 19751
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.902864259028642e-05,
+      "loss": 0.1109,
+      "step": 19752
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.901618929016189e-05,
+      "loss": 0.211,
+      "step": 19753
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.900373599003735e-05,
+      "loss": 0.0918,
+      "step": 19754
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.899128268991282e-05,
+      "loss": 0.1189,
+      "step": 19755
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.89788293897883e-05,
+      "loss": 0.1661,
+      "step": 19756
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.896637608966375e-05,
+      "loss": 0.3102,
+      "step": 19757
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.895392278953923e-05,
+      "loss": 0.2513,
+      "step": 19758
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.89414694894147e-05,
+      "loss": 0.212,
+      "step": 19759
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.892901618929016e-05,
+      "loss": 0.1358,
+      "step": 19760
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.891656288916563e-05,
+      "loss": 0.0722,
+      "step": 19761
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.890410958904108e-05,
+      "loss": 0.142,
+      "step": 19762
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 7.889165628891655e-05,
+      "loss": 0.3025,
+      "step": 19763
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.887920298879202e-05,
+      "loss": 0.1249,
+      "step": 19764
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.886674968866748e-05,
+      "loss": 0.1563,
+      "step": 19765
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.885429638854295e-05,
+      "loss": 0.1797,
+      "step": 19766
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.884184308841843e-05,
+      "loss": 0.0866,
+      "step": 19767
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.882938978829389e-05,
+      "loss": 0.1265,
+      "step": 19768
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.881693648816936e-05,
+      "loss": 0.1505,
+      "step": 19769
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.880448318804483e-05,
+      "loss": 0.2174,
+      "step": 19770
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.879202988792029e-05,
+      "loss": 0.203,
+      "step": 19771
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.877957658779576e-05,
+      "loss": 0.1378,
+      "step": 19772
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.876712328767122e-05,
+      "loss": 0.1015,
+      "step": 19773
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.87546699875467e-05,
+      "loss": 0.1825,
+      "step": 19774
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.874221668742217e-05,
+      "loss": 0.2129,
+      "step": 19775
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.872976338729763e-05,
+      "loss": 0.2134,
+      "step": 19776
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.87173100871731e-05,
+      "loss": 0.2117,
+      "step": 19777
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.870485678704856e-05,
+      "loss": 0.1367,
+      "step": 19778
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.869240348692402e-05,
+      "loss": 0.1314,
+      "step": 19779
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.867995018679949e-05,
+      "loss": 0.1302,
+      "step": 19780
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.866749688667496e-05,
+      "loss": 0.1715,
+      "step": 19781
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.865504358655042e-05,
+      "loss": 0.2065,
+      "step": 19782
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.86425902864259e-05,
+      "loss": 0.221,
+      "step": 19783
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.863013698630136e-05,
+      "loss": 0.1382,
+      "step": 19784
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.861768368617683e-05,
+      "loss": 0.1943,
+      "step": 19785
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.86052303860523e-05,
+      "loss": 0.1285,
+      "step": 19786
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.859277708592776e-05,
+      "loss": 0.0982,
+      "step": 19787
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.858032378580323e-05,
+      "loss": 0.1808,
+      "step": 19788
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 7.85678704856787e-05,
+      "loss": 0.1589,
+      "step": 19789
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.855541718555416e-05,
+      "loss": 0.1564,
+      "step": 19790
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.854296388542964e-05,
+      "loss": 0.1316,
+      "step": 19791
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.853051058530511e-05,
+      "loss": 0.1493,
+      "step": 19792
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.851805728518056e-05,
+      "loss": 0.1005,
+      "step": 19793
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.850560398505603e-05,
+      "loss": 0.1789,
+      "step": 19794
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.849315068493149e-05,
+      "loss": 0.2579,
+      "step": 19795
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.848069738480696e-05,
+      "loss": 0.1732,
+      "step": 19796
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.846824408468243e-05,
+      "loss": 0.1162,
+      "step": 19797
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.845579078455789e-05,
+      "loss": 0.11,
+      "step": 19798
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.844333748443336e-05,
+      "loss": 0.0711,
+      "step": 19799
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.843088418430884e-05,
+      "loss": 0.1697,
+      "step": 19800
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.84184308841843e-05,
+      "loss": 0.282,
+      "step": 19801
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.840597758405977e-05,
+      "loss": 0.1714,
+      "step": 19802
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.839352428393524e-05,
+      "loss": 0.2224,
+      "step": 19803
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.83810709838107e-05,
+      "loss": 0.1381,
+      "step": 19804
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.836861768368617e-05,
+      "loss": 0.0996,
+      "step": 19805
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.835616438356165e-05,
+      "loss": 0.236,
+      "step": 19806
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.83437110834371e-05,
+      "loss": 0.1979,
+      "step": 19807
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.833125778331258e-05,
+      "loss": 0.1113,
+      "step": 19808
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.831880448318802e-05,
+      "loss": 0.1189,
+      "step": 19809
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.83063511830635e-05,
+      "loss": 0.0939,
+      "step": 19810
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.829389788293897e-05,
+      "loss": 0.1657,
+      "step": 19811
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.828144458281443e-05,
+      "loss": 0.1411,
+      "step": 19812
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.82689912826899e-05,
+      "loss": 0.2265,
+      "step": 19813
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.825653798256537e-05,
+      "loss": 0.1388,
+      "step": 19814
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 7.824408468244083e-05,
+      "loss": 0.1058,
+      "step": 19815
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.82316313823163e-05,
+      "loss": 0.2029,
+      "step": 19816
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.821917808219178e-05,
+      "loss": 0.1391,
+      "step": 19817
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.820672478206724e-05,
+      "loss": 0.0989,
+      "step": 19818
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.819427148194271e-05,
+      "loss": 0.1584,
+      "step": 19819
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.818181818181818e-05,
+      "loss": 0.1741,
+      "step": 19820
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.816936488169364e-05,
+      "loss": 0.1295,
+      "step": 19821
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.815691158156912e-05,
+      "loss": 0.2274,
+      "step": 19822
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.814445828144459e-05,
+      "loss": 0.1057,
+      "step": 19823
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.813200498132005e-05,
+      "loss": 0.1179,
+      "step": 19824
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.81195516811955e-05,
+      "loss": 0.1775,
+      "step": 19825
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.810709838107097e-05,
+      "loss": 0.1702,
+      "step": 19826
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.809464508094644e-05,
+      "loss": 0.1825,
+      "step": 19827
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.808219178082191e-05,
+      "loss": 0.1589,
+      "step": 19828
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.806973848069737e-05,
+      "loss": 0.1181,
+      "step": 19829
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.805728518057284e-05,
+      "loss": 0.1308,
+      "step": 19830
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.804483188044832e-05,
+      "loss": 0.1591,
+      "step": 19831
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.803237858032377e-05,
+      "loss": 0.2189,
+      "step": 19832
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.801992528019925e-05,
+      "loss": 0.3395,
+      "step": 19833
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.800747198007472e-05,
+      "loss": 0.1286,
+      "step": 19834
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.799501867995018e-05,
+      "loss": 0.144,
+      "step": 19835
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.798256537982565e-05,
+      "loss": 0.0967,
+      "step": 19836
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.797011207970112e-05,
+      "loss": 0.1346,
+      "step": 19837
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.795765877957658e-05,
+      "loss": 0.2149,
+      "step": 19838
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.794520547945206e-05,
+      "loss": 0.1365,
+      "step": 19839
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.79327521793275e-05,
+      "loss": 0.1592,
+      "step": 19840
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 7.792029887920298e-05,
+      "loss": 0.142,
+      "step": 19841
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.790784557907845e-05,
+      "loss": 0.1309,
+      "step": 19842
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.789539227895391e-05,
+      "loss": 0.1113,
+      "step": 19843
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.788293897882938e-05,
+      "loss": 0.116,
+      "step": 19844
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.787048567870485e-05,
+      "loss": 0.2434,
+      "step": 19845
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.785803237858031e-05,
+      "loss": 0.1423,
+      "step": 19846
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.784557907845578e-05,
+      "loss": 0.163,
+      "step": 19847
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.783312577833126e-05,
+      "loss": 0.1077,
+      "step": 19848
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.782067247820672e-05,
+      "loss": 0.0986,
+      "step": 19849
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.780821917808219e-05,
+      "loss": 0.1178,
+      "step": 19850
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.779576587795766e-05,
+      "loss": 0.1385,
+      "step": 19851
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.778331257783312e-05,
+      "loss": 0.1258,
+      "step": 19852
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.77708592777086e-05,
+      "loss": 0.1142,
+      "step": 19853
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.775840597758405e-05,
+      "loss": 0.1492,
+      "step": 19854
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.774595267745953e-05,
+      "loss": 0.1663,
+      "step": 19855
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.773349937733498e-05,
+      "loss": 0.1614,
+      "step": 19856
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.772104607721044e-05,
+      "loss": 0.2697,
+      "step": 19857
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.770859277708592e-05,
+      "loss": 0.1595,
+      "step": 19858
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.769613947696139e-05,
+      "loss": 0.1205,
+      "step": 19859
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.768368617683685e-05,
+      "loss": 0.0932,
+      "step": 19860
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.767123287671232e-05,
+      "loss": 0.09,
+      "step": 19861
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.76587795765878e-05,
+      "loss": 0.1363,
+      "step": 19862
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.764632627646325e-05,
+      "loss": 0.1503,
+      "step": 19863
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.763387297633873e-05,
+      "loss": 0.1415,
+      "step": 19864
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.762141967621419e-05,
+      "loss": 0.1704,
+      "step": 19865
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.760896637608966e-05,
+      "loss": 0.093,
+      "step": 19866
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 7.759651307596513e-05,
+      "loss": 0.1272,
+      "step": 19867
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.758405977584059e-05,
+      "loss": 0.0944,
+      "step": 19868
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.757160647571606e-05,
+      "loss": 0.1204,
+      "step": 19869
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.755915317559154e-05,
+      "loss": 0.2238,
+      "step": 19870
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.7546699875467e-05,
+      "loss": 0.1825,
+      "step": 19871
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.753424657534245e-05,
+      "loss": 0.1449,
+      "step": 19872
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.752179327521793e-05,
+      "loss": 0.097,
+      "step": 19873
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.750933997509339e-05,
+      "loss": 0.125,
+      "step": 19874
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.749688667496886e-05,
+      "loss": 0.1678,
+      "step": 19875
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.748443337484432e-05,
+      "loss": 0.2255,
+      "step": 19876
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.747198007471979e-05,
+      "loss": 0.1813,
+      "step": 19877
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.745952677459526e-05,
+      "loss": 0.1458,
+      "step": 19878
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.744707347447072e-05,
+      "loss": 0.1161,
+      "step": 19879
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.74346201743462e-05,
+      "loss": 0.1269,
+      "step": 19880
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.742216687422167e-05,
+      "loss": 0.0998,
+      "step": 19881
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.740971357409713e-05,
+      "loss": 0.2432,
+      "step": 19882
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.73972602739726e-05,
+      "loss": 0.167,
+      "step": 19883
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.738480697384807e-05,
+      "loss": 0.1397,
+      "step": 19884
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.737235367372353e-05,
+      "loss": 0.0921,
+      "step": 19885
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.7359900373599e-05,
+      "loss": 0.1902,
+      "step": 19886
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.734744707347448e-05,
+      "loss": 0.1628,
+      "step": 19887
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.733499377334992e-05,
+      "loss": 0.1232,
+      "step": 19888
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.73225404732254e-05,
+      "loss": 0.168,
+      "step": 19889
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.731008717310085e-05,
+      "loss": 0.103,
+      "step": 19890
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.729763387297633e-05,
+      "loss": 0.1625,
+      "step": 19891
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.72851805728518e-05,
+      "loss": 0.1013,
+      "step": 19892
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 7.727272727272726e-05,
+      "loss": 0.1061,
+      "step": 19893
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.726027397260273e-05,
+      "loss": 0.103,
+      "step": 19894
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.72478206724782e-05,
+      "loss": 0.2351,
+      "step": 19895
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.723536737235366e-05,
+      "loss": 0.1766,
+      "step": 19896
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.722291407222914e-05,
+      "loss": 0.2068,
+      "step": 19897
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.721046077210461e-05,
+      "loss": 0.1298,
+      "step": 19898
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.719800747198007e-05,
+      "loss": 0.0835,
+      "step": 19899
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.718555417185554e-05,
+      "loss": 0.098,
+      "step": 19900
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.717310087173101e-05,
+      "loss": 0.2602,
+      "step": 19901
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.716064757160647e-05,
+      "loss": 0.1653,
+      "step": 19902
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.714819427148193e-05,
+      "loss": 0.1,
+      "step": 19903
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.713574097135739e-05,
+      "loss": 0.2106,
+      "step": 19904
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.712328767123286e-05,
+      "loss": 0.1215,
+      "step": 19905
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.711083437110834e-05,
+      "loss": 0.0867,
+      "step": 19906
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.70983810709838e-05,
+      "loss": 0.2096,
+      "step": 19907
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.708592777085927e-05,
+      "loss": 0.2047,
+      "step": 19908
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.707347447073474e-05,
+      "loss": 0.1262,
+      "step": 19909
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.70610211706102e-05,
+      "loss": 0.1379,
+      "step": 19910
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.704856787048567e-05,
+      "loss": 0.0842,
+      "step": 19911
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.703611457036115e-05,
+      "loss": 0.1304,
+      "step": 19912
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.70236612702366e-05,
+      "loss": 0.195,
+      "step": 19913
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.701120797011208e-05,
+      "loss": 0.2074,
+      "step": 19914
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.699875466998755e-05,
+      "loss": 0.156,
+      "step": 19915
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.698630136986301e-05,
+      "loss": 0.1293,
+      "step": 19916
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.697384806973848e-05,
+      "loss": 0.2141,
+      "step": 19917
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.696139476961395e-05,
+      "loss": 0.112,
+      "step": 19918
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 7.69489414694894e-05,
+      "loss": 0.1621,
+      "step": 19919
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.693648816936487e-05,
+      "loss": 0.2493,
+      "step": 19920
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.692403486924033e-05,
+      "loss": 0.1236,
+      "step": 19921
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.69115815691158e-05,
+      "loss": 0.1519,
+      "step": 19922
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.689912826899128e-05,
+      "loss": 0.0904,
+      "step": 19923
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.688667496886674e-05,
+      "loss": 0.0999,
+      "step": 19924
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.687422166874221e-05,
+      "loss": 0.0816,
+      "step": 19925
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.686176836861768e-05,
+      "loss": 0.3059,
+      "step": 19926
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.684931506849314e-05,
+      "loss": 0.1709,
+      "step": 19927
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.683686176836861e-05,
+      "loss": 0.1709,
+      "step": 19928
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.682440846824409e-05,
+      "loss": 0.1652,
+      "step": 19929
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.681195516811955e-05,
+      "loss": 0.146,
+      "step": 19930
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.679950186799502e-05,
+      "loss": 0.1866,
+      "step": 19931
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.678704856787049e-05,
+      "loss": 0.308,
+      "step": 19932
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.677459526774595e-05,
+      "loss": 0.2099,
+      "step": 19933
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.676214196762142e-05,
+      "loss": 0.1121,
+      "step": 19934
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.674968866749687e-05,
+      "loss": 0.1061,
+      "step": 19935
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.673723536737234e-05,
+      "loss": 0.17,
+      "step": 19936
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.672478206724781e-05,
+      "loss": 0.1018,
+      "step": 19937
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.671232876712327e-05,
+      "loss": 0.1656,
+      "step": 19938
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.669987546699875e-05,
+      "loss": 0.109,
+      "step": 19939
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.668742216687422e-05,
+      "loss": 0.2351,
+      "step": 19940
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.667496886674968e-05,
+      "loss": 0.1465,
+      "step": 19941
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.666251556662515e-05,
+      "loss": 0.1937,
+      "step": 19942
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.665006226650062e-05,
+      "loss": 0.1505,
+      "step": 19943
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.663760896637608e-05,
+      "loss": 0.0957,
+      "step": 19944
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 7.662515566625156e-05,
+      "loss": 0.1781,
+      "step": 19945
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.661270236612701e-05,
+      "loss": 0.1126,
+      "step": 19946
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.660024906600249e-05,
+      "loss": 0.1244,
+      "step": 19947
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.658779576587796e-05,
+      "loss": 0.1039,
+      "step": 19948
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.657534246575342e-05,
+      "loss": 0.129,
+      "step": 19949
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.656288916562888e-05,
+      "loss": 0.1678,
+      "step": 19950
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.655043586550435e-05,
+      "loss": 0.3085,
+      "step": 19951
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.653798256537981e-05,
+      "loss": 0.2486,
+      "step": 19952
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.652552926525528e-05,
+      "loss": 0.0779,
+      "step": 19953
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.651307596513076e-05,
+      "loss": 0.0911,
+      "step": 19954
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.650062266500621e-05,
+      "loss": 0.2388,
+      "step": 19955
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.648816936488169e-05,
+      "loss": 0.1225,
+      "step": 19956
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.647571606475715e-05,
+      "loss": 0.2682,
+      "step": 19957
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.646326276463262e-05,
+      "loss": 0.2048,
+      "step": 19958
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.645080946450809e-05,
+      "loss": 0.1028,
+      "step": 19959
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.643835616438355e-05,
+      "loss": 0.1859,
+      "step": 19960
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.642590286425902e-05,
+      "loss": 0.1762,
+      "step": 19961
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.64134495641345e-05,
+      "loss": 0.105,
+      "step": 19962
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.640099626400996e-05,
+      "loss": 0.3545,
+      "step": 19963
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.638854296388543e-05,
+      "loss": 0.1901,
+      "step": 19964
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.63760896637609e-05,
+      "loss": 0.1484,
+      "step": 19965
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.636363636363635e-05,
+      "loss": 0.1444,
+      "step": 19966
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.635118306351182e-05,
+      "loss": 0.1259,
+      "step": 19967
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.633872976338728e-05,
+      "loss": 0.1548,
+      "step": 19968
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.632627646326275e-05,
+      "loss": 0.1353,
+      "step": 19969
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.631382316313822e-05,
+      "loss": 0.169,
+      "step": 19970
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 7.630136986301368e-05,
+      "loss": 0.124,
+      "step": 19971
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.628891656288916e-05,
+      "loss": 0.113,
+      "step": 19972
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.627646326276463e-05,
+      "loss": 0.1552,
+      "step": 19973
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.626400996264009e-05,
+      "loss": 0.2049,
+      "step": 19974
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.625155666251556e-05,
+      "loss": 0.2199,
+      "step": 19975
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.623910336239103e-05,
+      "loss": 0.2209,
+      "step": 19976
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.622665006226649e-05,
+      "loss": 0.2052,
+      "step": 19977
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.621419676214197e-05,
+      "loss": 0.1022,
+      "step": 19978
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.620174346201744e-05,
+      "loss": 0.1226,
+      "step": 19979
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.61892901618929e-05,
+      "loss": 0.1682,
+      "step": 19980
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.617683686176837e-05,
+      "loss": 0.1107,
+      "step": 19981
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.616438356164382e-05,
+      "loss": 0.1662,
+      "step": 19982
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.615193026151929e-05,
+      "loss": 0.2003,
+      "step": 19983
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.613947696139476e-05,
+      "loss": 0.1225,
+      "step": 19984
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.612702366127022e-05,
+      "loss": 0.1504,
+      "step": 19985
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.611457036114569e-05,
+      "loss": 0.0806,
+      "step": 19986
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.610211706102117e-05,
+      "loss": 0.1465,
+      "step": 19987
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.608966376089662e-05,
+      "loss": 0.2329,
+      "step": 19988
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.60772104607721e-05,
+      "loss": 0.2565,
+      "step": 19989
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.606475716064757e-05,
+      "loss": 0.1473,
+      "step": 19990
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.605230386052303e-05,
+      "loss": 0.0933,
+      "step": 19991
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.60398505603985e-05,
+      "loss": 0.0966,
+      "step": 19992
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.602739726027398e-05,
+      "loss": 0.1589,
+      "step": 19993
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.601494396014943e-05,
+      "loss": 0.093,
+      "step": 19994
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.600249066002491e-05,
+      "loss": 0.229,
+      "step": 19995
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.599003735990038e-05,
+      "loss": 0.1332,
+      "step": 19996
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 7.597758405977583e-05,
+      "loss": 0.0925,
+      "step": 19997
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.59651307596513e-05,
+      "loss": 0.1559,
+      "step": 19998
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.595267745952676e-05,
+      "loss": 0.1828,
+      "step": 19999
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.594022415940223e-05,
+      "loss": 0.0946,
+      "step": 20000
+    },
+    {
+      "epoch": 7.67,
+      "eval_cer": 0.1016368496010301,
+      "eval_loss": 0.38376688957214355,
+      "eval_runtime": 47.4041,
+      "eval_samples_per_second": 36.98,
+      "eval_steps_per_second": 4.641,
+      "eval_wer": 0.16208082545141875,
+      "step": 20000
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.59277708592777e-05,
+      "loss": 0.2023,
+      "step": 20001
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.591531755915316e-05,
+      "loss": 0.213,
+      "step": 20002
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.590286425902863e-05,
+      "loss": 0.1014,
+      "step": 20003
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.589041095890411e-05,
+      "loss": 0.1176,
+      "step": 20004
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.587795765877957e-05,
+      "loss": 0.2247,
+      "step": 20005
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.586550435865504e-05,
+      "loss": 0.1084,
+      "step": 20006
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.585305105853051e-05,
+      "loss": 0.2071,
+      "step": 20007
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.584059775840597e-05,
+      "loss": 0.136,
+      "step": 20008
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.582814445828144e-05,
+      "loss": 0.1158,
+      "step": 20009
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.581569115815692e-05,
+      "loss": 0.1221,
+      "step": 20010
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.580323785803238e-05,
+      "loss": 0.2446,
+      "step": 20011
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.579078455790785e-05,
+      "loss": 0.2456,
+      "step": 20012
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.57783312577833e-05,
+      "loss": 0.102,
+      "step": 20013
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.576587795765877e-05,
+      "loss": 0.2556,
+      "step": 20014
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.575342465753424e-05,
+      "loss": 0.1592,
+      "step": 20015
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.57409713574097e-05,
+      "loss": 0.1499,
+      "step": 20016
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.572851805728517e-05,
+      "loss": 0.1255,
+      "step": 20017
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.571606475716064e-05,
+      "loss": 0.1113,
+      "step": 20018
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.57036114570361e-05,
+      "loss": 0.2289,
+      "step": 20019
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.569115815691158e-05,
+      "loss": 0.2803,
+      "step": 20020
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.567870485678705e-05,
+      "loss": 0.1308,
+      "step": 20021
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.566625155666251e-05,
+      "loss": 0.1992,
+      "step": 20022
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.565379825653798e-05,
+      "loss": 0.1764,
+      "step": 20023
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 7.564134495641345e-05,
+      "loss": 0.0852,
+      "step": 20024
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.562889165628891e-05,
+      "loss": 0.1279,
+      "step": 20025
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.561643835616439e-05,
+      "loss": 0.3162,
+      "step": 20026
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.560398505603984e-05,
+      "loss": 0.1485,
+      "step": 20027
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.559153175591532e-05,
+      "loss": 0.1898,
+      "step": 20028
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.557907845579078e-05,
+      "loss": 0.1271,
+      "step": 20029
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.556662515566624e-05,
+      "loss": 0.0943,
+      "step": 20030
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.555417185554171e-05,
+      "loss": 0.1386,
+      "step": 20031
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.554171855541718e-05,
+      "loss": 0.1606,
+      "step": 20032
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.552926525529264e-05,
+      "loss": 0.1448,
+      "step": 20033
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.551681195516811e-05,
+      "loss": 0.1359,
+      "step": 20034
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.550435865504359e-05,
+      "loss": 0.109,
+      "step": 20035
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.549190535491904e-05,
+      "loss": 0.0805,
+      "step": 20036
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.547945205479452e-05,
+      "loss": 0.1081,
+      "step": 20037
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.546699875466998e-05,
+      "loss": 0.1755,
+      "step": 20038
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.545454545454545e-05,
+      "loss": 0.163,
+      "step": 20039
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.544209215442092e-05,
+      "loss": 0.1636,
+      "step": 20040
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.542963885429638e-05,
+      "loss": 0.1702,
+      "step": 20041
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.541718555417185e-05,
+      "loss": 0.1139,
+      "step": 20042
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.540473225404733e-05,
+      "loss": 0.1363,
+      "step": 20043
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.539227895392279e-05,
+      "loss": 0.1921,
+      "step": 20044
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.537982565379824e-05,
+      "loss": 0.1741,
+      "step": 20045
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.53673723536737e-05,
+      "loss": 0.126,
+      "step": 20046
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.535491905354918e-05,
+      "loss": 0.1226,
+      "step": 20047
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.534246575342465e-05,
+      "loss": 0.1143,
+      "step": 20048
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.533001245330011e-05,
+      "loss": 0.0828,
+      "step": 20049
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 7.531755915317558e-05,
+      "loss": 0.1529,
+      "step": 20050
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.530510585305105e-05,
+      "loss": 0.2347,
+      "step": 20051
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.529265255292651e-05,
+      "loss": 0.2439,
+      "step": 20052
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.528019925280199e-05,
+      "loss": 0.125,
+      "step": 20053
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.526774595267746e-05,
+      "loss": 0.0837,
+      "step": 20054
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.525529265255292e-05,
+      "loss": 0.1299,
+      "step": 20055
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.524283935242839e-05,
+      "loss": 0.1888,
+      "step": 20056
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.523038605230386e-05,
+      "loss": 0.2469,
+      "step": 20057
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.521793275217932e-05,
+      "loss": 0.1819,
+      "step": 20058
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.52054794520548e-05,
+      "loss": 0.2002,
+      "step": 20059
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.519302615193024e-05,
+      "loss": 0.2016,
+      "step": 20060
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.518057285180571e-05,
+      "loss": 0.1031,
+      "step": 20061
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.516811955168119e-05,
+      "loss": 0.1515,
+      "step": 20062
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.515566625155665e-05,
+      "loss": 0.1863,
+      "step": 20063
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.514321295143212e-05,
+      "loss": 0.1833,
+      "step": 20064
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.513075965130759e-05,
+      "loss": 0.1818,
+      "step": 20065
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.511830635118305e-05,
+      "loss": 0.0911,
+      "step": 20066
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.510585305105852e-05,
+      "loss": 0.1067,
+      "step": 20067
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.5093399750934e-05,
+      "loss": 0.2271,
+      "step": 20068
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.508094645080945e-05,
+      "loss": 0.1805,
+      "step": 20069
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.506849315068493e-05,
+      "loss": 0.2668,
+      "step": 20070
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.50560398505604e-05,
+      "loss": 0.1096,
+      "step": 20071
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.504358655043586e-05,
+      "loss": 0.1182,
+      "step": 20072
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.503113325031133e-05,
+      "loss": 0.1376,
+      "step": 20073
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.50186799501868e-05,
+      "loss": 0.1081,
+      "step": 20074
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.500622665006226e-05,
+      "loss": 0.1367,
+      "step": 20075
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 7.499377334993772e-05,
+      "loss": 0.2133,
+      "step": 20076
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.49813200498132e-05,
+      "loss": 0.1607,
+      "step": 20077
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.496886674968867e-05,
+      "loss": 0.1153,
+      "step": 20078
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.495641344956413e-05,
+      "loss": 0.14,
+      "step": 20079
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.494396014943959e-05,
+      "loss": 0.1308,
+      "step": 20080
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.493150684931506e-05,
+      "loss": 0.1042,
+      "step": 20081
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.491905354919053e-05,
+      "loss": 0.3147,
+      "step": 20082
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.490660024906599e-05,
+      "loss": 0.1505,
+      "step": 20083
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.489414694894146e-05,
+      "loss": 0.1497,
+      "step": 20084
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.488169364881694e-05,
+      "loss": 0.1055,
+      "step": 20085
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.48692403486924e-05,
+      "loss": 0.1041,
+      "step": 20086
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.485678704856787e-05,
+      "loss": 0.2063,
+      "step": 20087
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.484433374844333e-05,
+      "loss": 0.2504,
+      "step": 20088
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.48318804483188e-05,
+      "loss": 0.1467,
+      "step": 20089
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.481942714819426e-05,
+      "loss": 0.1692,
+      "step": 20090
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.480697384806973e-05,
+      "loss": 0.1415,
+      "step": 20091
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.47945205479452e-05,
+      "loss": 0.1436,
+      "step": 20092
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.478206724782066e-05,
+      "loss": 0.1053,
+      "step": 20093
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.476961394769614e-05,
+      "loss": 0.2697,
+      "step": 20094
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.475716064757161e-05,
+      "loss": 0.1821,
+      "step": 20095
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.474470734744707e-05,
+      "loss": 0.1737,
+      "step": 20096
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.473225404732253e-05,
+      "loss": 0.1236,
+      "step": 20097
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.4719800747198e-05,
+      "loss": 0.1586,
+      "step": 20098
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.470734744707347e-05,
+      "loss": 0.1053,
+      "step": 20099
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.469489414694893e-05,
+      "loss": 0.1548,
+      "step": 20100
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.46824408468244e-05,
+      "loss": 0.2326,
+      "step": 20101
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 7.466998754669988e-05,
+      "loss": 0.1926,
+      "step": 20102
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.465753424657534e-05,
+      "loss": 0.1271,
+      "step": 20103
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.46450809464508e-05,
+      "loss": 0.0884,
+      "step": 20104
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.463262764632627e-05,
+      "loss": 0.0777,
+      "step": 20105
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.462017434620174e-05,
+      "loss": 0.1368,
+      "step": 20106
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.46077210460772e-05,
+      "loss": 0.1495,
+      "step": 20107
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.459526774595267e-05,
+      "loss": 0.148,
+      "step": 20108
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.458281444582815e-05,
+      "loss": 0.1212,
+      "step": 20109
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.45703611457036e-05,
+      "loss": 0.1621,
+      "step": 20110
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.455790784557906e-05,
+      "loss": 0.0654,
+      "step": 20111
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.454545454545454e-05,
+      "loss": 0.1211,
+      "step": 20112
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.453300124533001e-05,
+      "loss": 0.1789,
+      "step": 20113
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.452054794520547e-05,
+      "loss": 0.2398,
+      "step": 20114
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.450809464508094e-05,
+      "loss": 0.1643,
+      "step": 20115
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.44956413449564e-05,
+      "loss": 0.1839,
+      "step": 20116
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.448318804483187e-05,
+      "loss": 0.0862,
+      "step": 20117
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.447073474470735e-05,
+      "loss": 0.138,
+      "step": 20118
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.44582814445828e-05,
+      "loss": 0.1237,
+      "step": 20119
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.444582814445828e-05,
+      "loss": 0.2144,
+      "step": 20120
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.443337484433374e-05,
+      "loss": 0.1321,
+      "step": 20121
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.442092154420921e-05,
+      "loss": 0.1449,
+      "step": 20122
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.440846824408467e-05,
+      "loss": 0.2336,
+      "step": 20123
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.439601494396014e-05,
+      "loss": 0.1641,
+      "step": 20124
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.438356164383562e-05,
+      "loss": 0.0903,
+      "step": 20125
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.437110834371107e-05,
+      "loss": 0.2521,
+      "step": 20126
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.435865504358653e-05,
+      "loss": 0.1173,
+      "step": 20127
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 7.4346201743462e-05,
+      "loss": 0.1673,
+      "step": 20128
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.433374844333748e-05,
+      "loss": 0.1416,
+      "step": 20129
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.432129514321294e-05,
+      "loss": 0.109,
+      "step": 20130
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.430884184308841e-05,
+      "loss": 0.129,
+      "step": 20131
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.429638854296388e-05,
+      "loss": 0.1493,
+      "step": 20132
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.428393524283934e-05,
+      "loss": 0.1576,
+      "step": 20133
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.427148194271482e-05,
+      "loss": 0.1895,
+      "step": 20134
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.425902864259027e-05,
+      "loss": 0.106,
+      "step": 20135
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.424657534246575e-05,
+      "loss": 0.1352,
+      "step": 20136
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.42341220423412e-05,
+      "loss": 0.1619,
+      "step": 20137
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.422166874221668e-05,
+      "loss": 0.2509,
+      "step": 20138
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.420921544209215e-05,
+      "loss": 0.262,
+      "step": 20139
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.419676214196761e-05,
+      "loss": 0.1475,
+      "step": 20140
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.418430884184308e-05,
+      "loss": 0.1465,
+      "step": 20141
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.417185554171856e-05,
+      "loss": 0.0893,
+      "step": 20142
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.415940224159402e-05,
+      "loss": 0.1322,
+      "step": 20143
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.414694894146948e-05,
+      "loss": 0.0866,
+      "step": 20144
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.413449564134495e-05,
+      "loss": 0.3111,
+      "step": 20145
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.412204234122042e-05,
+      "loss": 0.1518,
+      "step": 20146
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.410958904109588e-05,
+      "loss": 0.1271,
+      "step": 20147
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.409713574097135e-05,
+      "loss": 0.0872,
+      "step": 20148
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.408468244084683e-05,
+      "loss": 0.1576,
+      "step": 20149
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.407222914072228e-05,
+      "loss": 0.1377,
+      "step": 20150
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.405977584059774e-05,
+      "loss": 0.3403,
+      "step": 20151
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.404732254047322e-05,
+      "loss": 0.2002,
+      "step": 20152
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.403486924034869e-05,
+      "loss": 0.2119,
+      "step": 20153
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 7.402241594022415e-05,
+      "loss": 0.0981,
+      "step": 20154
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.400996264009962e-05,
+      "loss": 0.075,
+      "step": 20155
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.39975093399751e-05,
+      "loss": 0.1133,
+      "step": 20156
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.398505603985055e-05,
+      "loss": 0.2002,
+      "step": 20157
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.397260273972601e-05,
+      "loss": 0.1344,
+      "step": 20158
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.396014943960148e-05,
+      "loss": 0.17,
+      "step": 20159
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.394769613947696e-05,
+      "loss": 0.1461,
+      "step": 20160
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.393524283935242e-05,
+      "loss": 0.1175,
+      "step": 20161
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.392278953922789e-05,
+      "loss": 0.1438,
+      "step": 20162
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.391033623910336e-05,
+      "loss": 0.1663,
+      "step": 20163
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.389788293897882e-05,
+      "loss": 0.2651,
+      "step": 20164
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.38854296388543e-05,
+      "loss": 0.1292,
+      "step": 20165
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.387297633872975e-05,
+      "loss": 0.1035,
+      "step": 20166
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.386052303860523e-05,
+      "loss": 0.0939,
+      "step": 20167
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.384806973848068e-05,
+      "loss": 0.103,
+      "step": 20168
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.383561643835616e-05,
+      "loss": 0.1261,
+      "step": 20169
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.382316313823163e-05,
+      "loss": 0.2018,
+      "step": 20170
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.381070983810709e-05,
+      "loss": 0.268,
+      "step": 20171
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.379825653798256e-05,
+      "loss": 0.1208,
+      "step": 20172
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.378580323785803e-05,
+      "loss": 0.1463,
+      "step": 20173
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.37733499377335e-05,
+      "loss": 0.1042,
+      "step": 20174
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.376089663760895e-05,
+      "loss": 0.1165,
+      "step": 20175
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.374844333748443e-05,
+      "loss": 0.1738,
+      "step": 20176
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.37359900373599e-05,
+      "loss": 0.1377,
+      "step": 20177
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.372353673723536e-05,
+      "loss": 0.2728,
+      "step": 20178
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.371108343711083e-05,
+      "loss": 0.1606,
+      "step": 20179
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 7.36986301369863e-05,
+      "loss": 0.1172,
+      "step": 20180
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.368617683686176e-05,
+      "loss": 0.1848,
+      "step": 20181
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.367372353673722e-05,
+      "loss": 0.1547,
+      "step": 20182
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.36612702366127e-05,
+      "loss": 0.156,
+      "step": 20183
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.364881693648817e-05,
+      "loss": 0.1114,
+      "step": 20184
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.363636363636363e-05,
+      "loss": 0.0947,
+      "step": 20185
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.36239103362391e-05,
+      "loss": 0.1404,
+      "step": 20186
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.361145703611457e-05,
+      "loss": 0.1459,
+      "step": 20187
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.359900373599003e-05,
+      "loss": 0.1467,
+      "step": 20188
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.35865504358655e-05,
+      "loss": 0.1824,
+      "step": 20189
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.357409713574096e-05,
+      "loss": 0.1842,
+      "step": 20190
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.356164383561644e-05,
+      "loss": 0.1426,
+      "step": 20191
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.35491905354919e-05,
+      "loss": 0.0954,
+      "step": 20192
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.353673723536737e-05,
+      "loss": 0.0683,
+      "step": 20193
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.352428393524284e-05,
+      "loss": 0.1169,
+      "step": 20194
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.35118306351183e-05,
+      "loss": 0.1702,
+      "step": 20195
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.349937733499377e-05,
+      "loss": 0.1509,
+      "step": 20196
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.348692403486923e-05,
+      "loss": 0.1155,
+      "step": 20197
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.34744707347447e-05,
+      "loss": 0.1433,
+      "step": 20198
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.346201743462016e-05,
+      "loss": 0.1074,
+      "step": 20199
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.344956413449564e-05,
+      "loss": 0.1623,
+      "step": 20200
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.343711083437111e-05,
+      "loss": 0.2354,
+      "step": 20201
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.342465753424657e-05,
+      "loss": 0.1932,
+      "step": 20202
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.341220423412204e-05,
+      "loss": 0.1605,
+      "step": 20203
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.33997509339975e-05,
+      "loss": 0.2055,
+      "step": 20204
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.338729763387297e-05,
+      "loss": 0.0978,
+      "step": 20205
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 7.337484433374843e-05,
+      "loss": 0.1654,
+      "step": 20206
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.33623910336239e-05,
+      "loss": 0.194,
+      "step": 20207
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.334993773349936e-05,
+      "loss": 0.1567,
+      "step": 20208
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.333748443337484e-05,
+      "loss": 0.1568,
+      "step": 20209
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.332503113325031e-05,
+      "loss": 0.1409,
+      "step": 20210
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.331257783312577e-05,
+      "loss": 0.1715,
+      "step": 20211
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.330012453300124e-05,
+      "loss": 0.1412,
+      "step": 20212
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.32876712328767e-05,
+      "loss": 0.2156,
+      "step": 20213
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.327521793275217e-05,
+      "loss": 0.1356,
+      "step": 20214
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.326276463262763e-05,
+      "loss": 0.098,
+      "step": 20215
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.32503113325031e-05,
+      "loss": 0.0939,
+      "step": 20216
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.323785803237858e-05,
+      "loss": 0.1512,
+      "step": 20217
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.322540473225404e-05,
+      "loss": 0.2407,
+      "step": 20218
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.321295143212951e-05,
+      "loss": 0.1061,
+      "step": 20219
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.320049813200498e-05,
+      "loss": 0.2235,
+      "step": 20220
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.318804483188044e-05,
+      "loss": 0.1185,
+      "step": 20221
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.31755915317559e-05,
+      "loss": 0.318,
+      "step": 20222
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.316313823163137e-05,
+      "loss": 0.1173,
+      "step": 20223
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.315068493150685e-05,
+      "loss": 0.1299,
+      "step": 20224
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.31382316313823e-05,
+      "loss": 0.1285,
+      "step": 20225
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.312577833125778e-05,
+      "loss": 0.372,
+      "step": 20226
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.311332503113325e-05,
+      "loss": 0.1347,
+      "step": 20227
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.310087173100871e-05,
+      "loss": 0.1653,
+      "step": 20228
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.308841843088417e-05,
+      "loss": 0.1316,
+      "step": 20229
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.307596513075964e-05,
+      "loss": 0.1216,
+      "step": 20230
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.306351183063511e-05,
+      "loss": 0.0619,
+      "step": 20231
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 7.305105853051057e-05,
+      "loss": 0.1763,
+      "step": 20232
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.303860523038605e-05,
+      "loss": 0.1772,
+      "step": 20233
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.302615193026152e-05,
+      "loss": 0.1917,
+      "step": 20234
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.301369863013698e-05,
+      "loss": 0.232,
+      "step": 20235
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.300124533001245e-05,
+      "loss": 0.119,
+      "step": 20236
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.298879202988791e-05,
+      "loss": 0.1536,
+      "step": 20237
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.297633872976338e-05,
+      "loss": 0.1418,
+      "step": 20238
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.296388542963884e-05,
+      "loss": 0.1326,
+      "step": 20239
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.295143212951431e-05,
+      "loss": 0.1902,
+      "step": 20240
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.293897882938979e-05,
+      "loss": 0.2078,
+      "step": 20241
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.292652552926525e-05,
+      "loss": 0.109,
+      "step": 20242
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.291407222914072e-05,
+      "loss": 0.108,
+      "step": 20243
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.290161892901619e-05,
+      "loss": 0.1995,
+      "step": 20244
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.288916562889165e-05,
+      "loss": 0.3361,
+      "step": 20245
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.287671232876711e-05,
+      "loss": 0.2214,
+      "step": 20246
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.286425902864258e-05,
+      "loss": 0.1811,
+      "step": 20247
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.285180572851806e-05,
+      "loss": 0.1343,
+      "step": 20248
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.283935242839351e-05,
+      "loss": 0.2242,
+      "step": 20249
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.282689912826899e-05,
+      "loss": 0.1083,
+      "step": 20250
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.281444582814446e-05,
+      "loss": 0.344,
+      "step": 20251
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.280199252801992e-05,
+      "loss": 0.1435,
+      "step": 20252
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.278953922789538e-05,
+      "loss": 0.1304,
+      "step": 20253
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.277708592777085e-05,
+      "loss": 0.152,
+      "step": 20254
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.276463262764632e-05,
+      "loss": 0.1717,
+      "step": 20255
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.275217932752178e-05,
+      "loss": 0.1066,
+      "step": 20256
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.273972602739726e-05,
+      "loss": 0.1573,
+      "step": 20257
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 0.1396,
+      "step": 20258
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.271481942714819e-05,
+      "loss": 0.1252,
+      "step": 20259
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.270236612702366e-05,
+      "loss": 0.1104,
+      "step": 20260
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.268991282689912e-05,
+      "loss": 0.155,
+      "step": 20261
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.267745952677459e-05,
+      "loss": 0.0978,
+      "step": 20262
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.266500622665005e-05,
+      "loss": 0.1882,
+      "step": 20263
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.265255292652552e-05,
+      "loss": 0.1365,
+      "step": 20264
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.2640099626401e-05,
+      "loss": 0.2116,
+      "step": 20265
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.262764632627646e-05,
+      "loss": 0.0937,
+      "step": 20266
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.261519302615193e-05,
+      "loss": 0.135,
+      "step": 20267
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.260273972602739e-05,
+      "loss": 0.1493,
+      "step": 20268
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.259028642590286e-05,
+      "loss": 0.1348,
+      "step": 20269
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.257783312577832e-05,
+      "loss": 0.1301,
+      "step": 20270
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.256537982565379e-05,
+      "loss": 0.1382,
+      "step": 20271
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.255292652552927e-05,
+      "loss": 0.2187,
+      "step": 20272
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.254047322540472e-05,
+      "loss": 0.0904,
+      "step": 20273
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.25280199252802e-05,
+      "loss": 0.1338,
+      "step": 20274
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.251556662515567e-05,
+      "loss": 0.1076,
+      "step": 20275
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.250311332503113e-05,
+      "loss": 0.1561,
+      "step": 20276
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.249066002490659e-05,
+      "loss": 0.1924,
+      "step": 20277
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.247820672478206e-05,
+      "loss": 0.1401,
+      "step": 20278
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.246575342465753e-05,
+      "loss": 0.1477,
+      "step": 20279
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.245330012453299e-05,
+      "loss": 0.171,
+      "step": 20280
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.244084682440847e-05,
+      "loss": 0.1033,
+      "step": 20281
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.242839352428394e-05,
+      "loss": 0.1091,
+      "step": 20282
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.24159402241594e-05,
+      "loss": 0.2352,
+      "step": 20283
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 7.240348692403486e-05,
+      "loss": 0.1924,
+      "step": 20284
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.239103362391033e-05,
+      "loss": 0.1196,
+      "step": 20285
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.23785803237858e-05,
+      "loss": 0.1271,
+      "step": 20286
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.236612702366126e-05,
+      "loss": 0.1215,
+      "step": 20287
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.235367372353673e-05,
+      "loss": 0.3252,
+      "step": 20288
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.234122042341219e-05,
+      "loss": 0.1626,
+      "step": 20289
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.232876712328767e-05,
+      "loss": 0.1145,
+      "step": 20290
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.231631382316314e-05,
+      "loss": 0.1828,
+      "step": 20291
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.23038605230386e-05,
+      "loss": 0.1338,
+      "step": 20292
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.229140722291406e-05,
+      "loss": 0.1625,
+      "step": 20293
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.227895392278953e-05,
+      "loss": 0.2018,
+      "step": 20294
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.2266500622665e-05,
+      "loss": 0.2243,
+      "step": 20295
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.225404732254046e-05,
+      "loss": 0.1611,
+      "step": 20296
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.224159402241593e-05,
+      "loss": 0.1648,
+      "step": 20297
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.22291407222914e-05,
+      "loss": 0.1796,
+      "step": 20298
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.221668742216687e-05,
+      "loss": 0.1235,
+      "step": 20299
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.220423412204233e-05,
+      "loss": 0.1161,
+      "step": 20300
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.21917808219178e-05,
+      "loss": 0.1745,
+      "step": 20301
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.217932752179327e-05,
+      "loss": 0.1615,
+      "step": 20302
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.216687422166873e-05,
+      "loss": 0.128,
+      "step": 20303
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.21544209215442e-05,
+      "loss": 0.2591,
+      "step": 20304
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.214196762141968e-05,
+      "loss": 0.102,
+      "step": 20305
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.212951432129513e-05,
+      "loss": 0.1335,
+      "step": 20306
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.211706102117061e-05,
+      "loss": 0.1584,
+      "step": 20307
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.210460772104607e-05,
+      "loss": 0.1528,
+      "step": 20308
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.209215442092154e-05,
+      "loss": 0.1597,
+      "step": 20309
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.2079701120797e-05,
+      "loss": 0.1383,
+      "step": 20310
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 7.206724782067247e-05,
+      "loss": 0.1535,
+      "step": 20311
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.205479452054794e-05,
+      "loss": 0.1154,
+      "step": 20312
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.20423412204234e-05,
+      "loss": 0.2333,
+      "step": 20313
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.202988792029888e-05,
+      "loss": 0.1752,
+      "step": 20314
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.201743462017433e-05,
+      "loss": 0.1073,
+      "step": 20315
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.200498132004981e-05,
+      "loss": 0.1204,
+      "step": 20316
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.199252801992527e-05,
+      "loss": 0.185,
+      "step": 20317
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.198007471980074e-05,
+      "loss": 0.0968,
+      "step": 20318
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.196762141967621e-05,
+      "loss": 0.0848,
+      "step": 20319
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.195516811955167e-05,
+      "loss": 0.235,
+      "step": 20320
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.194271481942714e-05,
+      "loss": 0.2349,
+      "step": 20321
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.193026151930262e-05,
+      "loss": 0.2151,
+      "step": 20322
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.191780821917808e-05,
+      "loss": 0.1106,
+      "step": 20323
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.190535491905353e-05,
+      "loss": 0.1635,
+      "step": 20324
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.189290161892901e-05,
+      "loss": 0.1505,
+      "step": 20325
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.188044831880448e-05,
+      "loss": 0.2469,
+      "step": 20326
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.186799501867994e-05,
+      "loss": 0.225,
+      "step": 20327
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.185554171855541e-05,
+      "loss": 0.1521,
+      "step": 20328
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.184308841843088e-05,
+      "loss": 0.1018,
+      "step": 20329
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.183063511830634e-05,
+      "loss": 0.1352,
+      "step": 20330
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.18181818181818e-05,
+      "loss": 0.1125,
+      "step": 20331
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.180572851805728e-05,
+      "loss": 0.2084,
+      "step": 20332
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.179327521793275e-05,
+      "loss": 0.157,
+      "step": 20333
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.178082191780821e-05,
+      "loss": 0.1616,
+      "step": 20334
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.176836861768368e-05,
+      "loss": 0.1547,
+      "step": 20335
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.175591531755915e-05,
+      "loss": 0.1928,
+      "step": 20336
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 7.174346201743461e-05,
+      "loss": 0.1246,
+      "step": 20337
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.173100871731009e-05,
+      "loss": 0.2947,
+      "step": 20338
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.171855541718554e-05,
+      "loss": 0.2819,
+      "step": 20339
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.170610211706102e-05,
+      "loss": 0.1302,
+      "step": 20340
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.169364881693648e-05,
+      "loss": 0.147,
+      "step": 20341
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.168119551681195e-05,
+      "loss": 0.1314,
+      "step": 20342
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.166874221668742e-05,
+      "loss": 0.0971,
+      "step": 20343
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.165628891656288e-05,
+      "loss": 0.1533,
+      "step": 20344
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.164383561643835e-05,
+      "loss": 0.2038,
+      "step": 20345
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.163138231631383e-05,
+      "loss": 0.114,
+      "step": 20346
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.161892901618929e-05,
+      "loss": 0.1404,
+      "step": 20347
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.160647571606474e-05,
+      "loss": 0.1239,
+      "step": 20348
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.159402241594022e-05,
+      "loss": 0.1257,
+      "step": 20349
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.158156911581569e-05,
+      "loss": 0.0924,
+      "step": 20350
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.156911581569115e-05,
+      "loss": 0.2818,
+      "step": 20351
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.155666251556662e-05,
+      "loss": 0.1434,
+      "step": 20352
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.15442092154421e-05,
+      "loss": 0.1993,
+      "step": 20353
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.153175591531755e-05,
+      "loss": 0.1008,
+      "step": 20354
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.151930261519301e-05,
+      "loss": 0.153,
+      "step": 20355
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.150684931506849e-05,
+      "loss": 0.0741,
+      "step": 20356
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.149439601494396e-05,
+      "loss": 0.2282,
+      "step": 20357
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.148194271481942e-05,
+      "loss": 0.2119,
+      "step": 20358
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.146948941469489e-05,
+      "loss": 0.2436,
+      "step": 20359
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.145703611457036e-05,
+      "loss": 0.1103,
+      "step": 20360
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.144458281444582e-05,
+      "loss": 0.1669,
+      "step": 20361
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.14321295143213e-05,
+      "loss": 0.2022,
+      "step": 20362
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 7.141967621419675e-05,
+      "loss": 0.1752,
+      "step": 20363
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.140722291407223e-05,
+      "loss": 0.2172,
+      "step": 20364
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.139476961394769e-05,
+      "loss": 0.1926,
+      "step": 20365
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.138231631382316e-05,
+      "loss": 0.1527,
+      "step": 20366
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.136986301369863e-05,
+      "loss": 0.1264,
+      "step": 20367
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.135740971357409e-05,
+      "loss": 0.1387,
+      "step": 20368
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.134495641344956e-05,
+      "loss": 0.1358,
+      "step": 20369
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.133250311332502e-05,
+      "loss": 0.2628,
+      "step": 20370
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.13200498132005e-05,
+      "loss": 0.2368,
+      "step": 20371
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.130759651307595e-05,
+      "loss": 0.1741,
+      "step": 20372
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.129514321295143e-05,
+      "loss": 0.1656,
+      "step": 20373
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.128268991282689e-05,
+      "loss": 0.2372,
+      "step": 20374
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.127023661270236e-05,
+      "loss": 0.297,
+      "step": 20375
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.125778331257783e-05,
+      "loss": 0.2686,
+      "step": 20376
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.124533001245329e-05,
+      "loss": 0.1346,
+      "step": 20377
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.123287671232876e-05,
+      "loss": 0.1159,
+      "step": 20378
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.122042341220422e-05,
+      "loss": 0.1412,
+      "step": 20379
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.12079701120797e-05,
+      "loss": 0.1807,
+      "step": 20380
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.119551681195515e-05,
+      "loss": 0.2989,
+      "step": 20381
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.118306351183063e-05,
+      "loss": 0.1949,
+      "step": 20382
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.11706102117061e-05,
+      "loss": 0.1694,
+      "step": 20383
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.115815691158156e-05,
+      "loss": 0.1542,
+      "step": 20384
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.114570361145703e-05,
+      "loss": 0.1396,
+      "step": 20385
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.113325031133249e-05,
+      "loss": 0.1999,
+      "step": 20386
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.112079701120796e-05,
+      "loss": 0.0776,
+      "step": 20387
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.110834371108342e-05,
+      "loss": 0.1093,
+      "step": 20388
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 7.10958904109589e-05,
+      "loss": 0.2093,
+      "step": 20389
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.108343711083437e-05,
+      "loss": 0.1537,
+      "step": 20390
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.107098381070983e-05,
+      "loss": 0.2462,
+      "step": 20391
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.10585305105853e-05,
+      "loss": 0.2744,
+      "step": 20392
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.104607721046077e-05,
+      "loss": 0.0892,
+      "step": 20393
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.103362391033623e-05,
+      "loss": 0.1792,
+      "step": 20394
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.102117061021169e-05,
+      "loss": 0.2521,
+      "step": 20395
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.100871731008716e-05,
+      "loss": 0.1586,
+      "step": 20396
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.099626400996264e-05,
+      "loss": 0.1203,
+      "step": 20397
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.09838107098381e-05,
+      "loss": 0.1533,
+      "step": 20398
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.097135740971357e-05,
+      "loss": 0.105,
+      "step": 20399
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.095890410958904e-05,
+      "loss": 0.1865,
+      "step": 20400
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.09464508094645e-05,
+      "loss": 0.2527,
+      "step": 20401
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.093399750933996e-05,
+      "loss": 0.1403,
+      "step": 20402
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.092154420921543e-05,
+      "loss": 0.1204,
+      "step": 20403
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.09090909090909e-05,
+      "loss": 0.1148,
+      "step": 20404
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.089663760896636e-05,
+      "loss": 0.1637,
+      "step": 20405
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.088418430884184e-05,
+      "loss": 0.0699,
+      "step": 20406
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.087173100871731e-05,
+      "loss": 0.2353,
+      "step": 20407
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.085927770859277e-05,
+      "loss": 0.0991,
+      "step": 20408
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.084682440846824e-05,
+      "loss": 0.1762,
+      "step": 20409
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.08343711083437e-05,
+      "loss": 0.1927,
+      "step": 20410
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.082191780821917e-05,
+      "loss": 0.1367,
+      "step": 20411
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.080946450809463e-05,
+      "loss": 0.0676,
+      "step": 20412
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.07970112079701e-05,
+      "loss": 0.1862,
+      "step": 20413
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.078455790784558e-05,
+      "loss": 0.1994,
+      "step": 20414
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 7.077210460772104e-05,
+      "loss": 0.1268,
+      "step": 20415
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.075965130759651e-05,
+      "loss": 0.1374,
+      "step": 20416
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.074719800747198e-05,
+      "loss": 0.078,
+      "step": 20417
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.073474470734744e-05,
+      "loss": 0.1156,
+      "step": 20418
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.07222914072229e-05,
+      "loss": 0.1124,
+      "step": 20419
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.070983810709837e-05,
+      "loss": 0.2558,
+      "step": 20420
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.069738480697385e-05,
+      "loss": 0.1815,
+      "step": 20421
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.06849315068493e-05,
+      "loss": 0.21,
+      "step": 20422
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.067247820672478e-05,
+      "loss": 0.1113,
+      "step": 20423
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.066002490660025e-05,
+      "loss": 0.1653,
+      "step": 20424
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.064757160647571e-05,
+      "loss": 0.1105,
+      "step": 20425
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.063511830635117e-05,
+      "loss": 0.2142,
+      "step": 20426
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.062266500622664e-05,
+      "loss": 0.1473,
+      "step": 20427
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.061021170610212e-05,
+      "loss": 0.1686,
+      "step": 20428
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.059775840597757e-05,
+      "loss": 0.1451,
+      "step": 20429
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.058530510585305e-05,
+      "loss": 0.0867,
+      "step": 20430
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.057285180572852e-05,
+      "loss": 0.1324,
+      "step": 20431
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.056039850560398e-05,
+      "loss": 0.3152,
+      "step": 20432
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.054794520547944e-05,
+      "loss": 0.1264,
+      "step": 20433
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.053549190535491e-05,
+      "loss": 0.1184,
+      "step": 20434
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.052303860523038e-05,
+      "loss": 0.1527,
+      "step": 20435
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.051058530510584e-05,
+      "loss": 0.1019,
+      "step": 20436
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.049813200498132e-05,
+      "loss": 0.0722,
+      "step": 20437
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.048567870485679e-05,
+      "loss": 0.2355,
+      "step": 20438
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.047322540473225e-05,
+      "loss": 0.1874,
+      "step": 20439
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.046077210460772e-05,
+      "loss": 0.1518,
+      "step": 20440
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 7.044831880448318e-05,
+      "loss": 0.132,
+      "step": 20441
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.043586550435865e-05,
+      "loss": 0.1297,
+      "step": 20442
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.042341220423411e-05,
+      "loss": 0.2219,
+      "step": 20443
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.041095890410958e-05,
+      "loss": 0.1192,
+      "step": 20444
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.039850560398506e-05,
+      "loss": 0.2814,
+      "step": 20445
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.038605230386052e-05,
+      "loss": 0.1539,
+      "step": 20446
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.037359900373599e-05,
+      "loss": 0.1743,
+      "step": 20447
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.036114570361146e-05,
+      "loss": 0.1458,
+      "step": 20448
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.034869240348692e-05,
+      "loss": 0.1285,
+      "step": 20449
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.033623910336238e-05,
+      "loss": 0.1078,
+      "step": 20450
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.032378580323785e-05,
+      "loss": 0.1588,
+      "step": 20451
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.031133250311332e-05,
+      "loss": 0.2043,
+      "step": 20452
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.029887920298878e-05,
+      "loss": 0.13,
+      "step": 20453
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.028642590286426e-05,
+      "loss": 0.1801,
+      "step": 20454
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.027397260273972e-05,
+      "loss": 0.1515,
+      "step": 20455
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.026151930261519e-05,
+      "loss": 0.1162,
+      "step": 20456
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.024906600249065e-05,
+      "loss": 0.214,
+      "step": 20457
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.023661270236612e-05,
+      "loss": 0.172,
+      "step": 20458
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.02241594022416e-05,
+      "loss": 0.111,
+      "step": 20459
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.021170610211705e-05,
+      "loss": 0.1256,
+      "step": 20460
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.019925280199253e-05,
+      "loss": 0.0828,
+      "step": 20461
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.018679950186798e-05,
+      "loss": 0.1561,
+      "step": 20462
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.017434620174346e-05,
+      "loss": 0.1668,
+      "step": 20463
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.016189290161893e-05,
+      "loss": 0.2268,
+      "step": 20464
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.014943960149439e-05,
+      "loss": 0.148,
+      "step": 20465
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.013698630136985e-05,
+      "loss": 0.1294,
+      "step": 20466
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 7.012453300124532e-05,
+      "loss": 0.1323,
+      "step": 20467
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.01120797011208e-05,
+      "loss": 0.2196,
+      "step": 20468
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.009962640099625e-05,
+      "loss": 0.1095,
+      "step": 20469
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.008717310087173e-05,
+      "loss": 0.2426,
+      "step": 20470
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.00747198007472e-05,
+      "loss": 0.1989,
+      "step": 20471
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.006226650062266e-05,
+      "loss": 0.2101,
+      "step": 20472
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.004981320049812e-05,
+      "loss": 0.1229,
+      "step": 20473
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.003735990037359e-05,
+      "loss": 0.1168,
+      "step": 20474
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.002490660024906e-05,
+      "loss": 0.1311,
+      "step": 20475
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7.001245330012452e-05,
+      "loss": 0.2561,
+      "step": 20476
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 7e-05,
+      "loss": 0.1489,
+      "step": 20477
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.998754669987547e-05,
+      "loss": 0.1778,
+      "step": 20478
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.997509339975093e-05,
+      "loss": 0.123,
+      "step": 20479
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.996264009962638e-05,
+      "loss": 0.1533,
+      "step": 20480
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.995018679950186e-05,
+      "loss": 0.1262,
+      "step": 20481
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.993773349937733e-05,
+      "loss": 0.3515,
+      "step": 20482
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.992528019925279e-05,
+      "loss": 0.1387,
+      "step": 20483
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.991282689912826e-05,
+      "loss": 0.2152,
+      "step": 20484
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.990037359900373e-05,
+      "loss": 0.1255,
+      "step": 20485
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.98879202988792e-05,
+      "loss": 0.2305,
+      "step": 20486
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.987546699875467e-05,
+      "loss": 0.0926,
+      "step": 20487
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.986301369863013e-05,
+      "loss": 0.2543,
+      "step": 20488
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.98505603985056e-05,
+      "loss": 0.2403,
+      "step": 20489
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.983810709838106e-05,
+      "loss": 0.1415,
+      "step": 20490
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.982565379825653e-05,
+      "loss": 0.1835,
+      "step": 20491
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.9813200498132e-05,
+      "loss": 0.1444,
+      "step": 20492
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 6.980074719800746e-05,
+      "loss": 0.1446,
+      "step": 20493
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.978829389788294e-05,
+      "loss": 0.1359,
+      "step": 20494
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.977584059775841e-05,
+      "loss": 0.237,
+      "step": 20495
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.976338729763387e-05,
+      "loss": 0.1054,
+      "step": 20496
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.975093399750933e-05,
+      "loss": 0.1541,
+      "step": 20497
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.97384806973848e-05,
+      "loss": 0.1723,
+      "step": 20498
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.972602739726027e-05,
+      "loss": 0.1419,
+      "step": 20499
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.971357409713573e-05,
+      "loss": 0.1455,
+      "step": 20500
+    },
+    {
+      "epoch": 7.86,
+      "eval_cer": 0.101956805899682,
+      "eval_loss": 0.37486258149147034,
+      "eval_runtime": 47.4996,
+      "eval_samples_per_second": 36.906,
+      "eval_steps_per_second": 4.632,
+      "eval_wer": 0.16518079377291034,
+      "step": 20500
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.97011207970112e-05,
+      "loss": 0.2402,
+      "step": 20501
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.968866749688668e-05,
+      "loss": 0.1753,
+      "step": 20502
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.967621419676214e-05,
+      "loss": 0.1234,
+      "step": 20503
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.96637608966376e-05,
+      "loss": 0.1406,
+      "step": 20504
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.965130759651307e-05,
+      "loss": 0.1839,
+      "step": 20505
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.963885429638854e-05,
+      "loss": 0.1353,
+      "step": 20506
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.9626400996264e-05,
+      "loss": 0.2195,
+      "step": 20507
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.961394769613947e-05,
+      "loss": 0.2077,
+      "step": 20508
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.960149439601494e-05,
+      "loss": 0.1053,
+      "step": 20509
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.95890410958904e-05,
+      "loss": 0.0992,
+      "step": 20510
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.957658779576588e-05,
+      "loss": 0.1373,
+      "step": 20511
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.956413449564134e-05,
+      "loss": 0.0936,
+      "step": 20512
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.955168119551681e-05,
+      "loss": 0.1049,
+      "step": 20513
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.953922789539227e-05,
+      "loss": 0.1475,
+      "step": 20514
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.952677459526774e-05,
+      "loss": 0.12,
+      "step": 20515
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.951432129514321e-05,
+      "loss": 0.1504,
+      "step": 20516
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.950186799501867e-05,
+      "loss": 0.0965,
+      "step": 20517
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.948941469489415e-05,
+      "loss": 0.164,
+      "step": 20518
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 6.947696139476962e-05,
+      "loss": 0.2458,
+      "step": 20519
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.946450809464508e-05,
+      "loss": 0.1943,
+      "step": 20520
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.945205479452054e-05,
+      "loss": 0.1528,
+      "step": 20521
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.943960149439601e-05,
+      "loss": 0.1357,
+      "step": 20522
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.942714819427148e-05,
+      "loss": 0.1148,
+      "step": 20523
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.941469489414694e-05,
+      "loss": 0.0799,
+      "step": 20524
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.940224159402241e-05,
+      "loss": 0.1175,
+      "step": 20525
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.938978829389789e-05,
+      "loss": 0.1682,
+      "step": 20526
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.937733499377335e-05,
+      "loss": 0.136,
+      "step": 20527
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.93648816936488e-05,
+      "loss": 0.1577,
+      "step": 20528
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.935242839352428e-05,
+      "loss": 0.1599,
+      "step": 20529
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.933997509339975e-05,
+      "loss": 0.139,
+      "step": 20530
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.932752179327521e-05,
+      "loss": 0.1268,
+      "step": 20531
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.931506849315068e-05,
+      "loss": 0.2133,
+      "step": 20532
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.930261519302615e-05,
+      "loss": 0.1595,
+      "step": 20533
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.929016189290161e-05,
+      "loss": 0.2767,
+      "step": 20534
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.927770859277707e-05,
+      "loss": 0.0993,
+      "step": 20535
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.926525529265255e-05,
+      "loss": 0.1715,
+      "step": 20536
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.925280199252802e-05,
+      "loss": 0.1025,
+      "step": 20537
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.924034869240348e-05,
+      "loss": 0.2646,
+      "step": 20538
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.922789539227895e-05,
+      "loss": 0.1707,
+      "step": 20539
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.921544209215441e-05,
+      "loss": 0.1801,
+      "step": 20540
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.920298879202988e-05,
+      "loss": 0.1751,
+      "step": 20541
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.919053549190535e-05,
+      "loss": 0.1663,
+      "step": 20542
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.917808219178081e-05,
+      "loss": 0.2544,
+      "step": 20543
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.916562889165629e-05,
+      "loss": 0.1039,
+      "step": 20544
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 6.915317559153175e-05,
+      "loss": 0.1451,
+      "step": 20545
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.914072229140722e-05,
+      "loss": 0.1183,
+      "step": 20546
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.912826899128268e-05,
+      "loss": 0.1662,
+      "step": 20547
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.911581569115815e-05,
+      "loss": 0.1016,
+      "step": 20548
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.910336239103362e-05,
+      "loss": 0.1043,
+      "step": 20549
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.909090909090908e-05,
+      "loss": 0.1131,
+      "step": 20550
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.907845579078454e-05,
+      "loss": 0.1568,
+      "step": 20551
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.906600249066001e-05,
+      "loss": 0.1627,
+      "step": 20552
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.905354919053549e-05,
+      "loss": 0.1105,
+      "step": 20553
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.904109589041095e-05,
+      "loss": 0.2165,
+      "step": 20554
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.902864259028642e-05,
+      "loss": 0.1481,
+      "step": 20555
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.901618929016189e-05,
+      "loss": 0.1205,
+      "step": 20556
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.900373599003735e-05,
+      "loss": 0.2536,
+      "step": 20557
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.899128268991282e-05,
+      "loss": 0.2164,
+      "step": 20558
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.897882938978828e-05,
+      "loss": 0.109,
+      "step": 20559
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.896637608966376e-05,
+      "loss": 0.1011,
+      "step": 20560
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.895392278953921e-05,
+      "loss": 0.1655,
+      "step": 20561
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.894146948941469e-05,
+      "loss": 0.2347,
+      "step": 20562
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.892901618929016e-05,
+      "loss": 0.2196,
+      "step": 20563
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.891656288916562e-05,
+      "loss": 0.1682,
+      "step": 20564
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.890410958904109e-05,
+      "loss": 0.1046,
+      "step": 20565
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.889165628891656e-05,
+      "loss": 0.183,
+      "step": 20566
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.887920298879202e-05,
+      "loss": 0.097,
+      "step": 20567
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.886674968866748e-05,
+      "loss": 0.0805,
+      "step": 20568
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.885429638854296e-05,
+      "loss": 0.2102,
+      "step": 20569
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.884184308841843e-05,
+      "loss": 0.2969,
+      "step": 20570
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 6.882938978829389e-05,
+      "loss": 0.1819,
+      "step": 20571
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.881693648816936e-05,
+      "loss": 0.141,
+      "step": 20572
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.880448318804483e-05,
+      "loss": 0.1311,
+      "step": 20573
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.879202988792029e-05,
+      "loss": 0.1352,
+      "step": 20574
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.877957658779575e-05,
+      "loss": 0.2304,
+      "step": 20575
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.876712328767122e-05,
+      "loss": 0.1386,
+      "step": 20576
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.87546699875467e-05,
+      "loss": 0.2328,
+      "step": 20577
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.874221668742216e-05,
+      "loss": 0.1478,
+      "step": 20578
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.872976338729763e-05,
+      "loss": 0.1795,
+      "step": 20579
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.87173100871731e-05,
+      "loss": 0.1239,
+      "step": 20580
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.870485678704856e-05,
+      "loss": 0.1376,
+      "step": 20581
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.869240348692402e-05,
+      "loss": 0.1025,
+      "step": 20582
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.867995018679949e-05,
+      "loss": 0.1556,
+      "step": 20583
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.866749688667497e-05,
+      "loss": 0.1172,
+      "step": 20584
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.865504358655042e-05,
+      "loss": 0.1374,
+      "step": 20585
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.86425902864259e-05,
+      "loss": 0.1217,
+      "step": 20586
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.863013698630137e-05,
+      "loss": 0.1562,
+      "step": 20587
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.861768368617683e-05,
+      "loss": 0.2324,
+      "step": 20588
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.86052303860523e-05,
+      "loss": 0.1851,
+      "step": 20589
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.859277708592776e-05,
+      "loss": 0.1894,
+      "step": 20590
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.858032378580323e-05,
+      "loss": 0.1602,
+      "step": 20591
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.856787048567869e-05,
+      "loss": 0.1345,
+      "step": 20592
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.855541718555417e-05,
+      "loss": 0.1282,
+      "step": 20593
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.854296388542964e-05,
+      "loss": 0.1246,
+      "step": 20594
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.85305105853051e-05,
+      "loss": 0.27,
+      "step": 20595
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.851805728518057e-05,
+      "loss": 0.2109,
+      "step": 20596
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.850560398505604e-05,
+      "loss": 0.2205,
+      "step": 20597
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 6.84931506849315e-05,
+      "loss": 0.1849,
+      "step": 20598
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.848069738480696e-05,
+      "loss": 0.1414,
+      "step": 20599
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.846824408468243e-05,
+      "loss": 0.1404,
+      "step": 20600
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.84557907845579e-05,
+      "loss": 0.3489,
+      "step": 20601
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.844333748443337e-05,
+      "loss": 0.1273,
+      "step": 20602
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.843088418430884e-05,
+      "loss": 0.17,
+      "step": 20603
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.841843088418431e-05,
+      "loss": 0.1496,
+      "step": 20604
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.840597758405977e-05,
+      "loss": 0.1622,
+      "step": 20605
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.839352428393523e-05,
+      "loss": 0.1937,
+      "step": 20606
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.83810709838107e-05,
+      "loss": 0.318,
+      "step": 20607
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.836861768368617e-05,
+      "loss": 0.197,
+      "step": 20608
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.835616438356163e-05,
+      "loss": 0.2042,
+      "step": 20609
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.834371108343711e-05,
+      "loss": 0.1291,
+      "step": 20610
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.833125778331258e-05,
+      "loss": 0.1773,
+      "step": 20611
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.831880448318804e-05,
+      "loss": 0.0898,
+      "step": 20612
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.830635118306351e-05,
+      "loss": 0.2069,
+      "step": 20613
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.829389788293897e-05,
+      "loss": 0.1399,
+      "step": 20614
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.828144458281444e-05,
+      "loss": 0.1979,
+      "step": 20615
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.82689912826899e-05,
+      "loss": 0.1173,
+      "step": 20616
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.825653798256538e-05,
+      "loss": 0.1233,
+      "step": 20617
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.824408468244085e-05,
+      "loss": 0.1394,
+      "step": 20618
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.823163138231631e-05,
+      "loss": 0.1251,
+      "step": 20619
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.821917808219178e-05,
+      "loss": 0.2124,
+      "step": 20620
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.820672478206724e-05,
+      "loss": 0.0972,
+      "step": 20621
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.819427148194271e-05,
+      "loss": 0.1481,
+      "step": 20622
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.818181818181817e-05,
+      "loss": 0.0976,
+      "step": 20623
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 6.816936488169364e-05,
+      "loss": 0.202,
+      "step": 20624
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.815691158156912e-05,
+      "loss": 0.1233,
+      "step": 20625
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.814445828144458e-05,
+      "loss": 0.1617,
+      "step": 20626
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.813200498132005e-05,
+      "loss": 0.1545,
+      "step": 20627
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.811955168119551e-05,
+      "loss": 0.1586,
+      "step": 20628
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.810709838107098e-05,
+      "loss": 0.176,
+      "step": 20629
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.809464508094644e-05,
+      "loss": 0.152,
+      "step": 20630
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.808219178082191e-05,
+      "loss": 0.0993,
+      "step": 20631
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.806973848069737e-05,
+      "loss": 0.2139,
+      "step": 20632
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.805728518057284e-05,
+      "loss": 0.2134,
+      "step": 20633
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.804483188044832e-05,
+      "loss": 0.1276,
+      "step": 20634
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.803237858032378e-05,
+      "loss": 0.1191,
+      "step": 20635
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.801992528019925e-05,
+      "loss": 0.0887,
+      "step": 20636
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.800747198007471e-05,
+      "loss": 0.1128,
+      "step": 20637
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.799501867995018e-05,
+      "loss": 0.263,
+      "step": 20638
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.798256537982564e-05,
+      "loss": 0.1643,
+      "step": 20639
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.797011207970111e-05,
+      "loss": 0.1295,
+      "step": 20640
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.795765877957659e-05,
+      "loss": 0.1447,
+      "step": 20641
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.794520547945204e-05,
+      "loss": 0.173,
+      "step": 20642
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.793275217932752e-05,
+      "loss": 0.1592,
+      "step": 20643
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.792029887920299e-05,
+      "loss": 0.1328,
+      "step": 20644
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.790784557907845e-05,
+      "loss": 0.2145,
+      "step": 20645
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.789539227895391e-05,
+      "loss": 0.0902,
+      "step": 20646
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.788293897882938e-05,
+      "loss": 0.1375,
+      "step": 20647
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.787048567870485e-05,
+      "loss": 0.1499,
+      "step": 20648
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.785803237858031e-05,
+      "loss": 0.111,
+      "step": 20649
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 6.784557907845579e-05,
+      "loss": 0.1047,
+      "step": 20650
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.783312577833126e-05,
+      "loss": 0.2291,
+      "step": 20651
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.782067247820672e-05,
+      "loss": 0.1939,
+      "step": 20652
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.780821917808218e-05,
+      "loss": 0.129,
+      "step": 20653
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.779576587795765e-05,
+      "loss": 0.2273,
+      "step": 20654
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.778331257783312e-05,
+      "loss": 0.1988,
+      "step": 20655
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.777085927770858e-05,
+      "loss": 0.1692,
+      "step": 20656
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.775840597758405e-05,
+      "loss": 0.2538,
+      "step": 20657
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.774595267745953e-05,
+      "loss": 0.2055,
+      "step": 20658
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.773349937733499e-05,
+      "loss": 0.1095,
+      "step": 20659
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.772104607721046e-05,
+      "loss": 0.0928,
+      "step": 20660
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.770859277708592e-05,
+      "loss": 0.1401,
+      "step": 20661
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.769613947696139e-05,
+      "loss": 0.1163,
+      "step": 20662
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.768368617683685e-05,
+      "loss": 0.2003,
+      "step": 20663
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.767123287671232e-05,
+      "loss": 0.1543,
+      "step": 20664
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.76587795765878e-05,
+      "loss": 0.1669,
+      "step": 20665
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.764632627646325e-05,
+      "loss": 0.1418,
+      "step": 20666
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.763387297633873e-05,
+      "loss": 0.0822,
+      "step": 20667
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.76214196762142e-05,
+      "loss": 0.1181,
+      "step": 20668
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.760896637608966e-05,
+      "loss": 0.1461,
+      "step": 20669
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.759651307596512e-05,
+      "loss": 0.282,
+      "step": 20670
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.758405977584059e-05,
+      "loss": 0.2745,
+      "step": 20671
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.757160647571606e-05,
+      "loss": 0.0847,
+      "step": 20672
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.755915317559152e-05,
+      "loss": 0.0835,
+      "step": 20673
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.7546699875467e-05,
+      "loss": 0.0851,
+      "step": 20674
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.753424657534247e-05,
+      "loss": 0.0909,
+      "step": 20675
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 6.752179327521793e-05,
+      "loss": 0.2385,
+      "step": 20676
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.750933997509339e-05,
+      "loss": 0.1594,
+      "step": 20677
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.749688667496886e-05,
+      "loss": 0.2025,
+      "step": 20678
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.748443337484433e-05,
+      "loss": 0.1533,
+      "step": 20679
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.747198007471979e-05,
+      "loss": 0.1829,
+      "step": 20680
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.745952677459526e-05,
+      "loss": 0.085,
+      "step": 20681
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.744707347447074e-05,
+      "loss": 0.2827,
+      "step": 20682
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.74346201743462e-05,
+      "loss": 0.1681,
+      "step": 20683
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.742216687422167e-05,
+      "loss": 0.1468,
+      "step": 20684
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.740971357409713e-05,
+      "loss": 0.1511,
+      "step": 20685
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.73972602739726e-05,
+      "loss": 0.1243,
+      "step": 20686
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.738480697384806e-05,
+      "loss": 0.222,
+      "step": 20687
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.737235367372353e-05,
+      "loss": 0.1478,
+      "step": 20688
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.7359900373599e-05,
+      "loss": 0.1716,
+      "step": 20689
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.734744707347446e-05,
+      "loss": 0.1358,
+      "step": 20690
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.733499377334994e-05,
+      "loss": 0.1269,
+      "step": 20691
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.73225404732254e-05,
+      "loss": 0.1665,
+      "step": 20692
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.731008717310087e-05,
+      "loss": 0.1897,
+      "step": 20693
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.729763387297633e-05,
+      "loss": 0.0919,
+      "step": 20694
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.72851805728518e-05,
+      "loss": 0.2262,
+      "step": 20695
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.727272727272727e-05,
+      "loss": 0.19,
+      "step": 20696
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.726027397260273e-05,
+      "loss": 0.2334,
+      "step": 20697
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.72478206724782e-05,
+      "loss": 0.139,
+      "step": 20698
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.723536737235368e-05,
+      "loss": 0.1113,
+      "step": 20699
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.722291407222914e-05,
+      "loss": 0.0991,
+      "step": 20700
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.72104607721046e-05,
+      "loss": 0.3458,
+      "step": 20701
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 6.719800747198007e-05,
+      "loss": 0.1807,
+      "step": 20702
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.718555417185554e-05,
+      "loss": 0.116,
+      "step": 20703
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.7173100871731e-05,
+      "loss": 0.1385,
+      "step": 20704
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.716064757160647e-05,
+      "loss": 0.2181,
+      "step": 20705
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.714819427148195e-05,
+      "loss": 0.2692,
+      "step": 20706
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.71357409713574e-05,
+      "loss": 0.1831,
+      "step": 20707
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.712328767123286e-05,
+      "loss": 0.2144,
+      "step": 20708
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.711083437110834e-05,
+      "loss": 0.2082,
+      "step": 20709
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.709838107098381e-05,
+      "loss": 0.1347,
+      "step": 20710
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.708592777085927e-05,
+      "loss": 0.2455,
+      "step": 20711
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.707347447073474e-05,
+      "loss": 0.1062,
+      "step": 20712
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.70610211706102e-05,
+      "loss": 0.1002,
+      "step": 20713
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.704856787048567e-05,
+      "loss": 0.1644,
+      "step": 20714
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.703611457036115e-05,
+      "loss": 0.1533,
+      "step": 20715
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.70236612702366e-05,
+      "loss": 0.1351,
+      "step": 20716
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.701120797011206e-05,
+      "loss": 0.0765,
+      "step": 20717
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.699875466998754e-05,
+      "loss": 0.1524,
+      "step": 20718
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.698630136986301e-05,
+      "loss": 0.0912,
+      "step": 20719
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.697384806973847e-05,
+      "loss": 0.2573,
+      "step": 20720
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.696139476961394e-05,
+      "loss": 0.1911,
+      "step": 20721
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.694894146948941e-05,
+      "loss": 0.1801,
+      "step": 20722
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.693648816936487e-05,
+      "loss": 0.2074,
+      "step": 20723
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.692403486924033e-05,
+      "loss": 0.1491,
+      "step": 20724
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.69115815691158e-05,
+      "loss": 0.1036,
+      "step": 20725
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.689912826899128e-05,
+      "loss": 0.2621,
+      "step": 20726
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.688667496886674e-05,
+      "loss": 0.2041,
+      "step": 20727
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 6.687422166874221e-05,
+      "loss": 0.2185,
+      "step": 20728
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.686176836861768e-05,
+      "loss": 0.1319,
+      "step": 20729
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.684931506849314e-05,
+      "loss": 0.1732,
+      "step": 20730
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.683686176836861e-05,
+      "loss": 0.1113,
+      "step": 20731
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.682440846824407e-05,
+      "loss": 0.1549,
+      "step": 20732
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.681195516811955e-05,
+      "loss": 0.2687,
+      "step": 20733
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.6799501867995e-05,
+      "loss": 0.1747,
+      "step": 20734
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.678704856787048e-05,
+      "loss": 0.1121,
+      "step": 20735
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.677459526774595e-05,
+      "loss": 0.1069,
+      "step": 20736
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.676214196762141e-05,
+      "loss": 0.1041,
+      "step": 20737
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.674968866749688e-05,
+      "loss": 0.2239,
+      "step": 20738
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.673723536737234e-05,
+      "loss": 0.2134,
+      "step": 20739
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.672478206724782e-05,
+      "loss": 0.2585,
+      "step": 20740
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.671232876712327e-05,
+      "loss": 0.2365,
+      "step": 20741
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.669987546699875e-05,
+      "loss": 0.1084,
+      "step": 20742
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.668742216687422e-05,
+      "loss": 0.1109,
+      "step": 20743
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.667496886674968e-05,
+      "loss": 0.2064,
+      "step": 20744
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.666251556662515e-05,
+      "loss": 0.2247,
+      "step": 20745
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.665006226650062e-05,
+      "loss": 0.1151,
+      "step": 20746
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.663760896637608e-05,
+      "loss": 0.1284,
+      "step": 20747
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.662515566625154e-05,
+      "loss": 0.1099,
+      "step": 20748
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.661270236612702e-05,
+      "loss": 0.1317,
+      "step": 20749
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.660024906600249e-05,
+      "loss": 0.1877,
+      "step": 20750
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.658779576587795e-05,
+      "loss": 0.1888,
+      "step": 20751
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.657534246575342e-05,
+      "loss": 0.2077,
+      "step": 20752
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.656288916562889e-05,
+      "loss": 0.1846,
+      "step": 20753
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 6.655043586550435e-05,
+      "loss": 0.0757,
+      "step": 20754
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.653798256537981e-05,
+      "loss": 0.0806,
+      "step": 20755
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.652552926525528e-05,
+      "loss": 0.1345,
+      "step": 20756
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.651307596513076e-05,
+      "loss": 0.2,
+      "step": 20757
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.650062266500622e-05,
+      "loss": 0.16,
+      "step": 20758
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.648816936488169e-05,
+      "loss": 0.1098,
+      "step": 20759
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.647571606475716e-05,
+      "loss": 0.1749,
+      "step": 20760
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.646326276463262e-05,
+      "loss": 0.115,
+      "step": 20761
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.645080946450809e-05,
+      "loss": 0.1296,
+      "step": 20762
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.643835616438355e-05,
+      "loss": 0.2729,
+      "step": 20763
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.642590286425903e-05,
+      "loss": 0.1587,
+      "step": 20764
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.641344956413448e-05,
+      "loss": 0.1507,
+      "step": 20765
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.640099626400996e-05,
+      "loss": 0.1825,
+      "step": 20766
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.638854296388543e-05,
+      "loss": 0.1545,
+      "step": 20767
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.637608966376089e-05,
+      "loss": 0.1283,
+      "step": 20768
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.636363636363636e-05,
+      "loss": 0.23,
+      "step": 20769
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.635118306351183e-05,
+      "loss": 0.2402,
+      "step": 20770
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.63387297633873e-05,
+      "loss": 0.2678,
+      "step": 20771
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.632627646326275e-05,
+      "loss": 0.2163,
+      "step": 20772
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.631382316313823e-05,
+      "loss": 0.1344,
+      "step": 20773
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.63013698630137e-05,
+      "loss": 0.1558,
+      "step": 20774
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.628891656288916e-05,
+      "loss": 0.0912,
+      "step": 20775
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.627646326276463e-05,
+      "loss": 0.2732,
+      "step": 20776
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.62640099626401e-05,
+      "loss": 0.2709,
+      "step": 20777
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.625155666251556e-05,
+      "loss": 0.1142,
+      "step": 20778
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.623910336239102e-05,
+      "loss": 0.1486,
+      "step": 20779
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 6.62266500622665e-05,
+      "loss": 0.0583,
+      "step": 20780
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.621419676214197e-05,
+      "loss": 0.124,
+      "step": 20781
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.620174346201743e-05,
+      "loss": 0.2286,
+      "step": 20782
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.61892901618929e-05,
+      "loss": 0.2139,
+      "step": 20783
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.617683686176837e-05,
+      "loss": 0.2085,
+      "step": 20784
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.616438356164383e-05,
+      "loss": 0.0807,
+      "step": 20785
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.61519302615193e-05,
+      "loss": 0.1382,
+      "step": 20786
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.613947696139476e-05,
+      "loss": 0.1398,
+      "step": 20787
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.612702366127023e-05,
+      "loss": 0.175,
+      "step": 20788
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.61145703611457e-05,
+      "loss": 0.1908,
+      "step": 20789
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.610211706102117e-05,
+      "loss": 0.1365,
+      "step": 20790
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.608966376089664e-05,
+      "loss": 0.2686,
+      "step": 20791
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.60772104607721e-05,
+      "loss": 0.1178,
+      "step": 20792
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.606475716064757e-05,
+      "loss": 0.0864,
+      "step": 20793
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.605230386052303e-05,
+      "loss": 0.1533,
+      "step": 20794
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.60398505603985e-05,
+      "loss": 0.2105,
+      "step": 20795
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.602739726027396e-05,
+      "loss": 0.2164,
+      "step": 20796
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.601494396014944e-05,
+      "loss": 0.1425,
+      "step": 20797
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.60024906600249e-05,
+      "loss": 0.1552,
+      "step": 20798
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.599003735990037e-05,
+      "loss": 0.1312,
+      "step": 20799
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.597758405977584e-05,
+      "loss": 0.1352,
+      "step": 20800
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.59651307596513e-05,
+      "loss": 0.2007,
+      "step": 20801
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.595267745952677e-05,
+      "loss": 0.118,
+      "step": 20802
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.594022415940223e-05,
+      "loss": 0.1974,
+      "step": 20803
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.59277708592777e-05,
+      "loss": 0.1391,
+      "step": 20804
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.591531755915316e-05,
+      "loss": 0.133,
+      "step": 20805
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 6.590286425902864e-05,
+      "loss": 0.1425,
+      "step": 20806
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.589041095890411e-05,
+      "loss": 0.2833,
+      "step": 20807
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.587795765877957e-05,
+      "loss": 0.1552,
+      "step": 20808
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.586550435865504e-05,
+      "loss": 0.2358,
+      "step": 20809
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.58530510585305e-05,
+      "loss": 0.1107,
+      "step": 20810
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.584059775840597e-05,
+      "loss": 0.091,
+      "step": 20811
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.582814445828143e-05,
+      "loss": 0.1139,
+      "step": 20812
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.58156911581569e-05,
+      "loss": 0.2098,
+      "step": 20813
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.580323785803238e-05,
+      "loss": 0.1921,
+      "step": 20814
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.579078455790784e-05,
+      "loss": 0.1734,
+      "step": 20815
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.577833125778331e-05,
+      "loss": 0.1542,
+      "step": 20816
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.576587795765878e-05,
+      "loss": 0.1567,
+      "step": 20817
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.575342465753424e-05,
+      "loss": 0.1157,
+      "step": 20818
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.57409713574097e-05,
+      "loss": 0.164,
+      "step": 20819
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.572851805728517e-05,
+      "loss": 0.2202,
+      "step": 20820
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.571606475716064e-05,
+      "loss": 0.1825,
+      "step": 20821
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.57036114570361e-05,
+      "loss": 0.1293,
+      "step": 20822
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.569115815691158e-05,
+      "loss": 0.11,
+      "step": 20823
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.567870485678705e-05,
+      "loss": 0.0872,
+      "step": 20824
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.566625155666251e-05,
+      "loss": 0.1259,
+      "step": 20825
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.565379825653797e-05,
+      "loss": 0.2573,
+      "step": 20826
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.564134495641344e-05,
+      "loss": 0.2076,
+      "step": 20827
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.562889165628891e-05,
+      "loss": 0.1342,
+      "step": 20828
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.561643835616437e-05,
+      "loss": 0.1817,
+      "step": 20829
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.560398505603985e-05,
+      "loss": 0.0924,
+      "step": 20830
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.559153175591532e-05,
+      "loss": 0.1557,
+      "step": 20831
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 6.557907845579078e-05,
+      "loss": 0.1999,
+      "step": 20832
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.556662515566625e-05,
+      "loss": 0.4012,
+      "step": 20833
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.555417185554171e-05,
+      "loss": 0.1368,
+      "step": 20834
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.554171855541718e-05,
+      "loss": 0.1456,
+      "step": 20835
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.552926525529264e-05,
+      "loss": 0.1069,
+      "step": 20836
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.551681195516811e-05,
+      "loss": 0.1262,
+      "step": 20837
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.550435865504359e-05,
+      "loss": 0.1719,
+      "step": 20838
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.549190535491905e-05,
+      "loss": 0.1904,
+      "step": 20839
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.547945205479452e-05,
+      "loss": 0.2002,
+      "step": 20840
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.546699875466999e-05,
+      "loss": 0.0912,
+      "step": 20841
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.545454545454545e-05,
+      "loss": 0.1356,
+      "step": 20842
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.544209215442091e-05,
+      "loss": 0.1311,
+      "step": 20843
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.542963885429638e-05,
+      "loss": 0.1233,
+      "step": 20844
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.542963885429638e-05,
+      "loss": 0.6159,
+      "step": 20845
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.541718555417185e-05,
+      "loss": 0.2784,
+      "step": 20846
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.540473225404731e-05,
+      "loss": 0.214,
+      "step": 20847
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.539227895392279e-05,
+      "loss": 0.0795,
+      "step": 20848
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.537982565379826e-05,
+      "loss": 0.0682,
+      "step": 20849
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.536737235367372e-05,
+      "loss": 0.0845,
+      "step": 20850
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.535491905354918e-05,
+      "loss": 0.2845,
+      "step": 20851
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.534246575342465e-05,
+      "loss": 0.1701,
+      "step": 20852
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.533001245330012e-05,
+      "loss": 0.1732,
+      "step": 20853
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.531755915317558e-05,
+      "loss": 0.1111,
+      "step": 20854
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.530510585305105e-05,
+      "loss": 0.104,
+      "step": 20855
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.529265255292653e-05,
+      "loss": 0.1317,
+      "step": 20856
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.528019925280199e-05,
+      "loss": 0.3354,
+      "step": 20857
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 6.526774595267745e-05,
+      "loss": 0.1946,
+      "step": 20858
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.525529265255292e-05,
+      "loss": 0.1395,
+      "step": 20859
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.524283935242839e-05,
+      "loss": 0.1442,
+      "step": 20860
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.523038605230385e-05,
+      "loss": 0.1289,
+      "step": 20861
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.521793275217932e-05,
+      "loss": 0.1502,
+      "step": 20862
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.52054794520548e-05,
+      "loss": 0.1328,
+      "step": 20863
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.519302615193026e-05,
+      "loss": 0.1761,
+      "step": 20864
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.518057285180573e-05,
+      "loss": 0.2521,
+      "step": 20865
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.516811955168119e-05,
+      "loss": 0.1257,
+      "step": 20866
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.515566625155666e-05,
+      "loss": 0.1467,
+      "step": 20867
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.514321295143212e-05,
+      "loss": 0.127,
+      "step": 20868
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.513075965130759e-05,
+      "loss": 0.0804,
+      "step": 20869
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.511830635118306e-05,
+      "loss": 0.2207,
+      "step": 20870
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.510585305105852e-05,
+      "loss": 0.1218,
+      "step": 20871
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.5093399750934e-05,
+      "loss": 0.1471,
+      "step": 20872
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.508094645080947e-05,
+      "loss": 0.1926,
+      "step": 20873
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.506849315068493e-05,
+      "loss": 0.1662,
+      "step": 20874
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.505603985056039e-05,
+      "loss": 0.1299,
+      "step": 20875
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.504358655043586e-05,
+      "loss": 0.1841,
+      "step": 20876
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.503113325031133e-05,
+      "loss": 0.0827,
+      "step": 20877
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.501867995018679e-05,
+      "loss": 0.1709,
+      "step": 20878
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.500622665006226e-05,
+      "loss": 0.2311,
+      "step": 20879
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.499377334993772e-05,
+      "loss": 0.1892,
+      "step": 20880
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.49813200498132e-05,
+      "loss": 0.1714,
+      "step": 20881
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.496886674968866e-05,
+      "loss": 0.1941,
+      "step": 20882
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.495641344956413e-05,
+      "loss": 0.1031,
+      "step": 20883
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.49439601494396e-05,
+      "loss": 0.1556,
+      "step": 20884
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 6.493150684931506e-05,
+      "loss": 0.2123,
+      "step": 20885
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.491905354919053e-05,
+      "loss": 0.1525,
+      "step": 20886
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.490660024906599e-05,
+      "loss": 0.0711,
+      "step": 20887
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.489414694894147e-05,
+      "loss": 0.1482,
+      "step": 20888
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.488169364881694e-05,
+      "loss": 0.1061,
+      "step": 20889
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.48692403486924e-05,
+      "loss": 0.165,
+      "step": 20890
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.485678704856786e-05,
+      "loss": 0.2184,
+      "step": 20891
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.484433374844333e-05,
+      "loss": 0.1459,
+      "step": 20892
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.48318804483188e-05,
+      "loss": 0.1943,
+      "step": 20893
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.481942714819426e-05,
+      "loss": 0.1561,
+      "step": 20894
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.480697384806973e-05,
+      "loss": 0.1037,
+      "step": 20895
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.47945205479452e-05,
+      "loss": 0.1407,
+      "step": 20896
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.478206724782067e-05,
+      "loss": 0.1804,
+      "step": 20897
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.476961394769612e-05,
+      "loss": 0.1969,
+      "step": 20898
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.47571606475716e-05,
+      "loss": 0.1926,
+      "step": 20899
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.474470734744707e-05,
+      "loss": 0.2035,
+      "step": 20900
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.473225404732253e-05,
+      "loss": 0.1177,
+      "step": 20901
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.4719800747198e-05,
+      "loss": 0.1048,
+      "step": 20902
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.470734744707347e-05,
+      "loss": 0.1238,
+      "step": 20903
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.469489414694893e-05,
+      "loss": 0.1924,
+      "step": 20904
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.468244084682439e-05,
+      "loss": 0.1129,
+      "step": 20905
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.466998754669987e-05,
+      "loss": 0.141,
+      "step": 20906
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.465753424657534e-05,
+      "loss": 0.1036,
+      "step": 20907
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.46450809464508e-05,
+      "loss": 0.1349,
+      "step": 20908
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.463262764632627e-05,
+      "loss": 0.1412,
+      "step": 20909
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.462017434620174e-05,
+      "loss": 0.1611,
+      "step": 20910
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 6.46077210460772e-05,
+      "loss": 0.206,
+      "step": 20911
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.459526774595267e-05,
+      "loss": 0.1365,
+      "step": 20912
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.458281444582813e-05,
+      "loss": 0.0966,
+      "step": 20913
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.45703611457036e-05,
+      "loss": 0.0898,
+      "step": 20914
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.455790784557907e-05,
+      "loss": 0.104,
+      "step": 20915
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.454545454545454e-05,
+      "loss": 0.1758,
+      "step": 20916
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.453300124533001e-05,
+      "loss": 0.1931,
+      "step": 20917
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.452054794520547e-05,
+      "loss": 0.1661,
+      "step": 20918
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.450809464508094e-05,
+      "loss": 0.0769,
+      "step": 20919
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.449564134495642e-05,
+      "loss": 0.0578,
+      "step": 20920
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.448318804483188e-05,
+      "loss": 0.0905,
+      "step": 20921
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.447073474470733e-05,
+      "loss": 0.149,
+      "step": 20922
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.445828144458281e-05,
+      "loss": 0.1553,
+      "step": 20923
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.444582814445828e-05,
+      "loss": 0.1871,
+      "step": 20924
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.443337484433374e-05,
+      "loss": 0.1275,
+      "step": 20925
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.442092154420921e-05,
+      "loss": 0.1282,
+      "step": 20926
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.440846824408468e-05,
+      "loss": 0.1771,
+      "step": 20927
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.439601494396014e-05,
+      "loss": 0.1517,
+      "step": 20928
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.43835616438356e-05,
+      "loss": 0.2731,
+      "step": 20929
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.437110834371108e-05,
+      "loss": 0.1374,
+      "step": 20930
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.435865504358655e-05,
+      "loss": 0.17,
+      "step": 20931
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.434620174346201e-05,
+      "loss": 0.1329,
+      "step": 20932
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.433374844333748e-05,
+      "loss": 0.0728,
+      "step": 20933
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.432129514321295e-05,
+      "loss": 0.067,
+      "step": 20934
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.430884184308841e-05,
+      "loss": 0.2713,
+      "step": 20935
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.429638854296388e-05,
+      "loss": 0.1721,
+      "step": 20936
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 6.428393524283934e-05,
+      "loss": 0.1236,
+      "step": 20937
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.427148194271482e-05,
+      "loss": 0.1021,
+      "step": 20938
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.425902864259028e-05,
+      "loss": 0.0686,
+      "step": 20939
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.424657534246575e-05,
+      "loss": 0.1218,
+      "step": 20940
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.423412204234122e-05,
+      "loss": 0.1971,
+      "step": 20941
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.422166874221668e-05,
+      "loss": 0.1254,
+      "step": 20942
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.420921544209215e-05,
+      "loss": 0.1144,
+      "step": 20943
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.419676214196763e-05,
+      "loss": 0.1199,
+      "step": 20944
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.418430884184308e-05,
+      "loss": 0.0737,
+      "step": 20945
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.417185554171854e-05,
+      "loss": 0.078,
+      "step": 20946
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.415940224159402e-05,
+      "loss": 0.3453,
+      "step": 20947
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.414694894146949e-05,
+      "loss": 0.1426,
+      "step": 20948
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.413449564134495e-05,
+      "loss": 0.1212,
+      "step": 20949
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.412204234122042e-05,
+      "loss": 0.172,
+      "step": 20950
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.41095890410959e-05,
+      "loss": 0.1139,
+      "step": 20951
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.409713574097135e-05,
+      "loss": 0.1445,
+      "step": 20952
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.408468244084681e-05,
+      "loss": 0.1243,
+      "step": 20953
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.407222914072229e-05,
+      "loss": 0.1999,
+      "step": 20954
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.405977584059776e-05,
+      "loss": 0.1591,
+      "step": 20955
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.404732254047322e-05,
+      "loss": 0.1269,
+      "step": 20956
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.403486924034869e-05,
+      "loss": 0.1904,
+      "step": 20957
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.402241594022416e-05,
+      "loss": 0.1463,
+      "step": 20958
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.400996264009962e-05,
+      "loss": 0.1193,
+      "step": 20959
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.399750933997508e-05,
+      "loss": 0.1615,
+      "step": 20960
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.398505603985055e-05,
+      "loss": 0.1177,
+      "step": 20961
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.397260273972603e-05,
+      "loss": 0.1127,
+      "step": 20962
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 6.396014943960149e-05,
+      "loss": 0.0938,
+      "step": 20963
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.394769613947696e-05,
+      "loss": 0.0669,
+      "step": 20964
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.393524283935243e-05,
+      "loss": 0.1207,
+      "step": 20965
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.392278953922789e-05,
+      "loss": 0.2249,
+      "step": 20966
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.391033623910336e-05,
+      "loss": 0.2304,
+      "step": 20967
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.389788293897882e-05,
+      "loss": 0.0993,
+      "step": 20968
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.38854296388543e-05,
+      "loss": 0.1677,
+      "step": 20969
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.387297633872975e-05,
+      "loss": 0.0816,
+      "step": 20970
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.386052303860523e-05,
+      "loss": 0.1084,
+      "step": 20971
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.384806973848069e-05,
+      "loss": 0.3078,
+      "step": 20972
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.383561643835616e-05,
+      "loss": 0.2018,
+      "step": 20973
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.382316313823163e-05,
+      "loss": 0.2068,
+      "step": 20974
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.381070983810709e-05,
+      "loss": 0.1051,
+      "step": 20975
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.379825653798255e-05,
+      "loss": 0.2885,
+      "step": 20976
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.378580323785802e-05,
+      "loss": 0.1592,
+      "step": 20977
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.37733499377335e-05,
+      "loss": 0.0893,
+      "step": 20978
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.376089663760895e-05,
+      "loss": 0.1858,
+      "step": 20979
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.374844333748443e-05,
+      "loss": 0.131,
+      "step": 20980
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.37359900373599e-05,
+      "loss": 0.1931,
+      "step": 20981
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.372353673723536e-05,
+      "loss": 0.0965,
+      "step": 20982
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.371108343711083e-05,
+      "loss": 0.066,
+      "step": 20983
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.369863013698629e-05,
+      "loss": 0.1348,
+      "step": 20984
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.368617683686176e-05,
+      "loss": 0.2064,
+      "step": 20985
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.367372353673722e-05,
+      "loss": 0.1395,
+      "step": 20986
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.36612702366127e-05,
+      "loss": 0.0763,
+      "step": 20987
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.364881693648817e-05,
+      "loss": 0.104,
+      "step": 20988
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 6.363636363636363e-05,
+      "loss": 0.0947,
+      "step": 20989
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.36239103362391e-05,
+      "loss": 0.1718,
+      "step": 20990
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.361145703611457e-05,
+      "loss": 0.1335,
+      "step": 20991
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.359900373599003e-05,
+      "loss": 0.1885,
+      "step": 20992
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.358655043586549e-05,
+      "loss": 0.1327,
+      "step": 20993
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.357409713574096e-05,
+      "loss": 0.13,
+      "step": 20994
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.356164383561644e-05,
+      "loss": 0.152,
+      "step": 20995
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.35491905354919e-05,
+      "loss": 0.1127,
+      "step": 20996
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.353673723536737e-05,
+      "loss": 0.2097,
+      "step": 20997
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.352428393524284e-05,
+      "loss": 0.2159,
+      "step": 20998
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.35118306351183e-05,
+      "loss": 0.1262,
+      "step": 20999
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.349937733499376e-05,
+      "loss": 0.1303,
+      "step": 21000
+    },
+    {
+      "epoch": 8.05,
+      "eval_cer": 0.10107107321926762,
+      "eval_loss": 0.40741202235221863,
+      "eval_runtime": 47.6896,
+      "eval_samples_per_second": 36.759,
+      "eval_steps_per_second": 4.613,
+      "eval_wer": 0.16146988278951893,
+      "step": 21000
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.348692403486923e-05,
+      "loss": 0.1662,
+      "step": 21001
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.34744707347447e-05,
+      "loss": 0.1246,
+      "step": 21002
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.346201743462016e-05,
+      "loss": 0.1234,
+      "step": 21003
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.344956413449564e-05,
+      "loss": 0.197,
+      "step": 21004
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.343711083437111e-05,
+      "loss": 0.1535,
+      "step": 21005
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.342465753424657e-05,
+      "loss": 0.0745,
+      "step": 21006
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.341220423412203e-05,
+      "loss": 0.1726,
+      "step": 21007
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.33997509339975e-05,
+      "loss": 0.2114,
+      "step": 21008
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.338729763387297e-05,
+      "loss": 0.108,
+      "step": 21009
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.337484433374843e-05,
+      "loss": 0.2245,
+      "step": 21010
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.33623910336239e-05,
+      "loss": 0.1169,
+      "step": 21011
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.334993773349938e-05,
+      "loss": 0.1511,
+      "step": 21012
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.333748443337484e-05,
+      "loss": 0.1551,
+      "step": 21013
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.332503113325031e-05,
+      "loss": 0.1557,
+      "step": 21014
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 6.331257783312577e-05,
+      "loss": 0.0618,
+      "step": 21015
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.330012453300124e-05,
+      "loss": 0.1552,
+      "step": 21016
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.32876712328767e-05,
+      "loss": 0.2311,
+      "step": 21017
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.327521793275217e-05,
+      "loss": 0.0927,
+      "step": 21018
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.326276463262765e-05,
+      "loss": 0.1115,
+      "step": 21019
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.32503113325031e-05,
+      "loss": 0.1448,
+      "step": 21020
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.323785803237858e-05,
+      "loss": 0.1148,
+      "step": 21021
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.322540473225405e-05,
+      "loss": 0.1807,
+      "step": 21022
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.321295143212951e-05,
+      "loss": 0.1438,
+      "step": 21023
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.320049813200497e-05,
+      "loss": 0.1539,
+      "step": 21024
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.318804483188044e-05,
+      "loss": 0.0861,
+      "step": 21025
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.317559153175591e-05,
+      "loss": 0.1018,
+      "step": 21026
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.316313823163137e-05,
+      "loss": 0.0819,
+      "step": 21027
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.315068493150685e-05,
+      "loss": 0.0699,
+      "step": 21028
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.313823163138232e-05,
+      "loss": 0.159,
+      "step": 21029
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.312577833125778e-05,
+      "loss": 0.158,
+      "step": 21030
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.311332503113324e-05,
+      "loss": 0.1172,
+      "step": 21031
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.310087173100871e-05,
+      "loss": 0.1764,
+      "step": 21032
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.308841843088418e-05,
+      "loss": 0.2018,
+      "step": 21033
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.307596513075964e-05,
+      "loss": 0.0927,
+      "step": 21034
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.306351183063511e-05,
+      "loss": 0.2517,
+      "step": 21035
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.305105853051059e-05,
+      "loss": 0.1386,
+      "step": 21036
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.303860523038605e-05,
+      "loss": 0.1293,
+      "step": 21037
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.302615193026152e-05,
+      "loss": 0.0972,
+      "step": 21038
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.301369863013698e-05,
+      "loss": 0.0956,
+      "step": 21039
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.300124533001245e-05,
+      "loss": 0.0715,
+      "step": 21040
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 6.298879202988791e-05,
+      "loss": 0.2238,
+      "step": 21041
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.297633872976338e-05,
+      "loss": 0.1549,
+      "step": 21042
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.296388542963886e-05,
+      "loss": 0.1444,
+      "step": 21043
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.295143212951432e-05,
+      "loss": 0.1156,
+      "step": 21044
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.293897882938979e-05,
+      "loss": 0.0938,
+      "step": 21045
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.292652552926525e-05,
+      "loss": 0.0721,
+      "step": 21046
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.291407222914072e-05,
+      "loss": 0.2,
+      "step": 21047
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.290161892901618e-05,
+      "loss": 0.175,
+      "step": 21048
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.288916562889165e-05,
+      "loss": 0.1383,
+      "step": 21049
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.287671232876712e-05,
+      "loss": 0.1356,
+      "step": 21050
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.286425902864258e-05,
+      "loss": 0.1059,
+      "step": 21051
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.285180572851806e-05,
+      "loss": 0.0925,
+      "step": 21052
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.283935242839352e-05,
+      "loss": 0.1792,
+      "step": 21053
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.282689912826899e-05,
+      "loss": 0.1411,
+      "step": 21054
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.281444582814445e-05,
+      "loss": 0.1147,
+      "step": 21055
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.280199252801992e-05,
+      "loss": 0.0944,
+      "step": 21056
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.278953922789538e-05,
+      "loss": 0.1353,
+      "step": 21057
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.277708592777085e-05,
+      "loss": 0.1316,
+      "step": 21058
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.276463262764632e-05,
+      "loss": 0.1176,
+      "step": 21059
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.275217932752178e-05,
+      "loss": 0.1758,
+      "step": 21060
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.273972602739726e-05,
+      "loss": 0.2093,
+      "step": 21061
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.272727272727272e-05,
+      "loss": 0.1661,
+      "step": 21062
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.271481942714819e-05,
+      "loss": 0.1054,
+      "step": 21063
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.270236612702365e-05,
+      "loss": 0.077,
+      "step": 21064
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.268991282689912e-05,
+      "loss": 0.118,
+      "step": 21065
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.267745952677459e-05,
+      "loss": 0.1801,
+      "step": 21066
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 6.266500622665005e-05,
+      "loss": 0.1364,
+      "step": 21067
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.265255292652552e-05,
+      "loss": 0.176,
+      "step": 21068
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.2640099626401e-05,
+      "loss": 0.1199,
+      "step": 21069
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.262764632627646e-05,
+      "loss": 0.2116,
+      "step": 21070
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.261519302615192e-05,
+      "loss": 0.0891,
+      "step": 21071
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.260273972602739e-05,
+      "loss": 0.1804,
+      "step": 21072
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.259028642590286e-05,
+      "loss": 0.1471,
+      "step": 21073
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.257783312577832e-05,
+      "loss": 0.1212,
+      "step": 21074
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.25653798256538e-05,
+      "loss": 0.0968,
+      "step": 21075
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.255292652552927e-05,
+      "loss": 0.1606,
+      "step": 21076
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.254047322540473e-05,
+      "loss": 0.0825,
+      "step": 21077
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.252801992528018e-05,
+      "loss": 0.1035,
+      "step": 21078
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.251556662515566e-05,
+      "loss": 0.1947,
+      "step": 21079
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.250311332503113e-05,
+      "loss": 0.137,
+      "step": 21080
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.249066002490659e-05,
+      "loss": 0.1605,
+      "step": 21081
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.247820672478206e-05,
+      "loss": 0.0782,
+      "step": 21082
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.246575342465753e-05,
+      "loss": 0.0802,
+      "step": 21083
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.2453300124533e-05,
+      "loss": 0.134,
+      "step": 21084
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.244084682440847e-05,
+      "loss": 0.2002,
+      "step": 21085
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.242839352428393e-05,
+      "loss": 0.1364,
+      "step": 21086
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.24159402241594e-05,
+      "loss": 0.1423,
+      "step": 21087
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.240348692403486e-05,
+      "loss": 0.1411,
+      "step": 21088
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.239103362391033e-05,
+      "loss": 0.0628,
+      "step": 21089
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.23785803237858e-05,
+      "loss": 0.0952,
+      "step": 21090
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.236612702366126e-05,
+      "loss": 0.2475,
+      "step": 21091
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.235367372353673e-05,
+      "loss": 0.1487,
+      "step": 21092
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 6.234122042341221e-05,
+      "loss": 0.1407,
+      "step": 21093
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.232876712328767e-05,
+      "loss": 0.1108,
+      "step": 21094
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.231631382316313e-05,
+      "loss": 0.1217,
+      "step": 21095
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.23038605230386e-05,
+      "loss": 0.0935,
+      "step": 21096
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.229140722291407e-05,
+      "loss": 0.2059,
+      "step": 21097
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.227895392278953e-05,
+      "loss": 0.2278,
+      "step": 21098
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.2266500622665e-05,
+      "loss": 0.1676,
+      "step": 21099
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.225404732254048e-05,
+      "loss": 0.1538,
+      "step": 21100
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.224159402241593e-05,
+      "loss": 0.0823,
+      "step": 21101
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.22291407222914e-05,
+      "loss": 0.169,
+      "step": 21102
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.221668742216687e-05,
+      "loss": 0.1072,
+      "step": 21103
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.220423412204234e-05,
+      "loss": 0.1791,
+      "step": 21104
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.21917808219178e-05,
+      "loss": 0.1185,
+      "step": 21105
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.217932752179327e-05,
+      "loss": 0.1331,
+      "step": 21106
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.216687422166874e-05,
+      "loss": 0.0763,
+      "step": 21107
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.21544209215442e-05,
+      "loss": 0.0579,
+      "step": 21108
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.214196762141968e-05,
+      "loss": 0.1183,
+      "step": 21109
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.212951432129514e-05,
+      "loss": 0.2216,
+      "step": 21110
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.211706102117061e-05,
+      "loss": 0.177,
+      "step": 21111
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.210460772104607e-05,
+      "loss": 0.136,
+      "step": 21112
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.209215442092154e-05,
+      "loss": 0.137,
+      "step": 21113
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.207970112079701e-05,
+      "loss": 0.1094,
+      "step": 21114
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.206724782067247e-05,
+      "loss": 0.127,
+      "step": 21115
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.205479452054794e-05,
+      "loss": 0.1817,
+      "step": 21116
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.20423412204234e-05,
+      "loss": 0.2506,
+      "step": 21117
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.202988792029888e-05,
+      "loss": 0.1297,
+      "step": 21118
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 6.201743462017434e-05,
+      "loss": 0.0971,
+      "step": 21119
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.200498132004981e-05,
+      "loss": 0.0983,
+      "step": 21120
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.199252801992528e-05,
+      "loss": 0.0789,
+      "step": 21121
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.198007471980074e-05,
+      "loss": 0.211,
+      "step": 21122
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.196762141967621e-05,
+      "loss": 0.1598,
+      "step": 21123
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.195516811955169e-05,
+      "loss": 0.11,
+      "step": 21124
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.194271481942714e-05,
+      "loss": 0.1472,
+      "step": 21125
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.19302615193026e-05,
+      "loss": 0.1217,
+      "step": 21126
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.191780821917808e-05,
+      "loss": 0.0898,
+      "step": 21127
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.190535491905355e-05,
+      "loss": 0.129,
+      "step": 21128
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.189290161892901e-05,
+      "loss": 0.2345,
+      "step": 21129
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.188044831880448e-05,
+      "loss": 0.1094,
+      "step": 21130
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.186799501867995e-05,
+      "loss": 0.0868,
+      "step": 21131
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.185554171855541e-05,
+      "loss": 0.0834,
+      "step": 21132
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.184308841843087e-05,
+      "loss": 0.0877,
+      "step": 21133
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.183063511830634e-05,
+      "loss": 0.1365,
+      "step": 21134
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.181818181818182e-05,
+      "loss": 0.1967,
+      "step": 21135
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.180572851805728e-05,
+      "loss": 0.2112,
+      "step": 21136
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.179327521793275e-05,
+      "loss": 0.1018,
+      "step": 21137
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.178082191780821e-05,
+      "loss": 0.1528,
+      "step": 21138
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.176836861768368e-05,
+      "loss": 0.0922,
+      "step": 21139
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.175591531755915e-05,
+      "loss": 0.086,
+      "step": 21140
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.174346201743461e-05,
+      "loss": 0.276,
+      "step": 21141
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.173100871731007e-05,
+      "loss": 0.1434,
+      "step": 21142
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.171855541718555e-05,
+      "loss": 0.1372,
+      "step": 21143
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.170610211706102e-05,
+      "loss": 0.0786,
+      "step": 21144
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 6.169364881693648e-05,
+      "loss": 0.0942,
+      "step": 21145
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.168119551681195e-05,
+      "loss": 0.0764,
+      "step": 21146
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.166874221668742e-05,
+      "loss": 0.2219,
+      "step": 21147
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.165628891656288e-05,
+      "loss": 0.1805,
+      "step": 21148
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.164383561643834e-05,
+      "loss": 0.1187,
+      "step": 21149
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.163138231631381e-05,
+      "loss": 0.1019,
+      "step": 21150
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.161892901618929e-05,
+      "loss": 0.0956,
+      "step": 21151
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.160647571606475e-05,
+      "loss": 0.0838,
+      "step": 21152
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.159402241594022e-05,
+      "loss": 0.1213,
+      "step": 21153
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.158156911581569e-05,
+      "loss": 0.3278,
+      "step": 21154
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.156911581569115e-05,
+      "loss": 0.1759,
+      "step": 21155
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.155666251556662e-05,
+      "loss": 0.1254,
+      "step": 21156
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.154420921544208e-05,
+      "loss": 0.0923,
+      "step": 21157
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.153175591531755e-05,
+      "loss": 0.2322,
+      "step": 21158
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.151930261519301e-05,
+      "loss": 0.1124,
+      "step": 21159
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.150684931506849e-05,
+      "loss": 0.1596,
+      "step": 21160
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.149439601494396e-05,
+      "loss": 0.1456,
+      "step": 21161
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.148194271481942e-05,
+      "loss": 0.1224,
+      "step": 21162
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.146948941469489e-05,
+      "loss": 0.0873,
+      "step": 21163
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.145703611457036e-05,
+      "loss": 0.1011,
+      "step": 21164
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.144458281444582e-05,
+      "loss": 0.1283,
+      "step": 21165
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.143212951432128e-05,
+      "loss": 0.1895,
+      "step": 21166
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.141967621419676e-05,
+      "loss": 0.1562,
+      "step": 21167
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.140722291407223e-05,
+      "loss": 0.1506,
+      "step": 21168
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.139476961394769e-05,
+      "loss": 0.144,
+      "step": 21169
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.138231631382316e-05,
+      "loss": 0.2456,
+      "step": 21170
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.136986301369863e-05,
+      "loss": 0.1436,
+      "step": 21171
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 6.135740971357409e-05,
+      "loss": 0.1945,
+      "step": 21172
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.134495641344955e-05,
+      "loss": 0.1955,
+      "step": 21173
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.133250311332502e-05,
+      "loss": 0.1598,
+      "step": 21174
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.13200498132005e-05,
+      "loss": 0.1689,
+      "step": 21175
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.130759651307596e-05,
+      "loss": 0.1159,
+      "step": 21176
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.129514321295143e-05,
+      "loss": 0.1102,
+      "step": 21177
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.12826899128269e-05,
+      "loss": 0.1361,
+      "step": 21178
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.127023661270236e-05,
+      "loss": 0.1822,
+      "step": 21179
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.125778331257782e-05,
+      "loss": 0.1645,
+      "step": 21180
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.124533001245329e-05,
+      "loss": 0.1539,
+      "step": 21181
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.123287671232876e-05,
+      "loss": 0.1309,
+      "step": 21182
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.122042341220422e-05,
+      "loss": 0.1004,
+      "step": 21183
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.12079701120797e-05,
+      "loss": 0.0957,
+      "step": 21184
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.119551681195517e-05,
+      "loss": 0.1526,
+      "step": 21185
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.118306351183063e-05,
+      "loss": 0.1125,
+      "step": 21186
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.11706102117061e-05,
+      "loss": 0.2188,
+      "step": 21187
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.115815691158156e-05,
+      "loss": 0.1255,
+      "step": 21188
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.114570361145703e-05,
+      "loss": 0.1315,
+      "step": 21189
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.113325031133249e-05,
+      "loss": 0.1265,
+      "step": 21190
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.112079701120796e-05,
+      "loss": 0.1382,
+      "step": 21191
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.110834371108344e-05,
+      "loss": 0.1594,
+      "step": 21192
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.10958904109589e-05,
+      "loss": 0.2616,
+      "step": 21193
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.108343711083437e-05,
+      "loss": 0.1556,
+      "step": 21194
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.107098381070984e-05,
+      "loss": 0.0936,
+      "step": 21195
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.10585305105853e-05,
+      "loss": 0.0802,
+      "step": 21196
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.104607721046076e-05,
+      "loss": 0.2331,
+      "step": 21197
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 6.103362391033623e-05,
+      "loss": 0.2977,
+      "step": 21198
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.10211706102117e-05,
+      "loss": 0.1305,
+      "step": 21199
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.1008717310087165e-05,
+      "loss": 0.1712,
+      "step": 21200
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.099626400996264e-05,
+      "loss": 0.107,
+      "step": 21201
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.0983810709838104e-05,
+      "loss": 0.1105,
+      "step": 21202
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.097135740971357e-05,
+      "loss": 0.0895,
+      "step": 21203
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.0958904109589036e-05,
+      "loss": 0.2837,
+      "step": 21204
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.09464508094645e-05,
+      "loss": 0.123,
+      "step": 21205
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.093399750933997e-05,
+      "loss": 0.1224,
+      "step": 21206
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.0921544209215433e-05,
+      "loss": 0.1014,
+      "step": 21207
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.0909090909090906e-05,
+      "loss": 0.1798,
+      "step": 21208
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.089663760896637e-05,
+      "loss": 0.1462,
+      "step": 21209
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.088418430884184e-05,
+      "loss": 0.2008,
+      "step": 21210
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.087173100871731e-05,
+      "loss": 0.188,
+      "step": 21211
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.085927770859277e-05,
+      "loss": 0.0992,
+      "step": 21212
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.0846824408468236e-05,
+      "loss": 0.1101,
+      "step": 21213
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.08343711083437e-05,
+      "loss": 0.1245,
+      "step": 21214
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.0821917808219175e-05,
+      "loss": 0.1281,
+      "step": 21215
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.080946450809464e-05,
+      "loss": 0.2672,
+      "step": 21216
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.0797011207970107e-05,
+      "loss": 0.1796,
+      "step": 21217
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.078455790784558e-05,
+      "loss": 0.1135,
+      "step": 21218
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.077210460772104e-05,
+      "loss": 0.099,
+      "step": 21219
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.0759651307596504e-05,
+      "loss": 0.1007,
+      "step": 21220
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.074719800747197e-05,
+      "loss": 0.0809,
+      "step": 21221
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.073474470734744e-05,
+      "loss": 0.124,
+      "step": 21222
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.072229140722291e-05,
+      "loss": 0.1519,
+      "step": 21223
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 6.0709838107098375e-05,
+      "loss": 0.1217,
+      "step": 21224
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.069738480697385e-05,
+      "loss": 0.0965,
+      "step": 21225
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.0684931506849314e-05,
+      "loss": 0.0884,
+      "step": 21226
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.067247820672477e-05,
+      "loss": 0.2151,
+      "step": 21227
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.066002490660024e-05,
+      "loss": 0.0958,
+      "step": 21228
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.064757160647571e-05,
+      "loss": 0.2586,
+      "step": 21229
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.063511830635118e-05,
+      "loss": 0.197,
+      "step": 21230
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.062266500622664e-05,
+      "loss": 0.117,
+      "step": 21231
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.0610211706102116e-05,
+      "loss": 0.1512,
+      "step": 21232
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.059775840597758e-05,
+      "loss": 0.1669,
+      "step": 21233
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.058530510585305e-05,
+      "loss": 0.0701,
+      "step": 21234
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.057285180572851e-05,
+      "loss": 0.1821,
+      "step": 21235
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.056039850560398e-05,
+      "loss": 0.0984,
+      "step": 21236
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.0547945205479446e-05,
+      "loss": 0.1088,
+      "step": 21237
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.053549190535491e-05,
+      "loss": 0.1063,
+      "step": 21238
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.0523038605230384e-05,
+      "loss": 0.1459,
+      "step": 21239
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.051058530510585e-05,
+      "loss": 0.1281,
+      "step": 21240
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.0498132004981316e-05,
+      "loss": 0.1695,
+      "step": 21241
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.048567870485678e-05,
+      "loss": 0.107,
+      "step": 21242
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.047322540473225e-05,
+      "loss": 0.1077,
+      "step": 21243
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.0460772104607714e-05,
+      "loss": 0.1172,
+      "step": 21244
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.044831880448318e-05,
+      "loss": 0.0943,
+      "step": 21245
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.0435865504358646e-05,
+      "loss": 0.0815,
+      "step": 21246
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.042341220423412e-05,
+      "loss": 0.2477,
+      "step": 21247
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.0410958904109585e-05,
+      "loss": 0.1156,
+      "step": 21248
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.039850560398505e-05,
+      "loss": 0.1382,
+      "step": 21249
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 6.038605230386052e-05,
+      "loss": 0.1236,
+      "step": 21250
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.037359900373598e-05,
+      "loss": 0.1387,
+      "step": 21251
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.036114570361145e-05,
+      "loss": 0.1477,
+      "step": 21252
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.0348692403486914e-05,
+      "loss": 0.1493,
+      "step": 21253
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.033623910336239e-05,
+      "loss": 0.234,
+      "step": 21254
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.032378580323785e-05,
+      "loss": 0.1592,
+      "step": 21255
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.031133250311332e-05,
+      "loss": 0.0928,
+      "step": 21256
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.029887920298879e-05,
+      "loss": 0.1492,
+      "step": 21257
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.028642590286426e-05,
+      "loss": 0.0951,
+      "step": 21258
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.027397260273972e-05,
+      "loss": 0.0949,
+      "step": 21259
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.026151930261518e-05,
+      "loss": 0.1616,
+      "step": 21260
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.0249066002490656e-05,
+      "loss": 0.1508,
+      "step": 21261
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.023661270236612e-05,
+      "loss": 0.128,
+      "step": 21262
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.022415940224159e-05,
+      "loss": 0.0731,
+      "step": 21263
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.021170610211706e-05,
+      "loss": 0.1335,
+      "step": 21264
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.0199252801992526e-05,
+      "loss": 0.1113,
+      "step": 21265
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.018679950186799e-05,
+      "loss": 0.1419,
+      "step": 21266
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.017434620174345e-05,
+      "loss": 0.1573,
+      "step": 21267
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.0161892901618924e-05,
+      "loss": 0.1024,
+      "step": 21268
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.014943960149439e-05,
+      "loss": 0.1032,
+      "step": 21269
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.0136986301369856e-05,
+      "loss": 0.1173,
+      "step": 21270
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.012453300124533e-05,
+      "loss": 0.1023,
+      "step": 21271
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.0112079701120794e-05,
+      "loss": 0.154,
+      "step": 21272
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.009962640099626e-05,
+      "loss": 0.1336,
+      "step": 21273
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.008717310087172e-05,
+      "loss": 0.0983,
+      "step": 21274
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.007471980074719e-05,
+      "loss": 0.1302,
+      "step": 21275
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 6.006226650062266e-05,
+      "loss": 0.0605,
+      "step": 21276
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.0049813200498124e-05,
+      "loss": 0.1512,
+      "step": 21277
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.00373599003736e-05,
+      "loss": 0.1332,
+      "step": 21278
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.002490660024906e-05,
+      "loss": 0.2566,
+      "step": 21279
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 6.001245330012453e-05,
+      "loss": 0.2542,
+      "step": 21280
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 0.1281,
+      "step": 21281
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.998754669987546e-05,
+      "loss": 0.1079,
+      "step": 21282
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.997509339975093e-05,
+      "loss": 0.1317,
+      "step": 21283
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.996264009962639e-05,
+      "loss": 0.0807,
+      "step": 21284
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.9950186799501865e-05,
+      "loss": 0.1557,
+      "step": 21285
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.993773349937733e-05,
+      "loss": 0.1607,
+      "step": 21286
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.99252801992528e-05,
+      "loss": 0.2211,
+      "step": 21287
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.991282689912826e-05,
+      "loss": 0.0957,
+      "step": 21288
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.9900373599003736e-05,
+      "loss": 0.0934,
+      "step": 21289
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.9887920298879195e-05,
+      "loss": 0.2095,
+      "step": 21290
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.987546699875466e-05,
+      "loss": 0.1955,
+      "step": 21291
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.986301369863013e-05,
+      "loss": 0.1406,
+      "step": 21292
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.98505603985056e-05,
+      "loss": 0.1615,
+      "step": 21293
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.9838107098381066e-05,
+      "loss": 0.1891,
+      "step": 21294
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.982565379825653e-05,
+      "loss": 0.1308,
+      "step": 21295
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.9813200498132004e-05,
+      "loss": 0.1958,
+      "step": 21296
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.980074719800747e-05,
+      "loss": 0.3759,
+      "step": 21297
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.978829389788293e-05,
+      "loss": 0.1683,
+      "step": 21298
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.9775840597758395e-05,
+      "loss": 0.109,
+      "step": 21299
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.976338729763387e-05,
+      "loss": 0.2077,
+      "step": 21300
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.9750933997509334e-05,
+      "loss": 0.1533,
+      "step": 21301
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 5.97384806973848e-05,
+      "loss": 0.1063,
+      "step": 21302
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.972602739726027e-05,
+      "loss": 0.0805,
+      "step": 21303
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.971357409713574e-05,
+      "loss": 0.2323,
+      "step": 21304
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.9701120797011205e-05,
+      "loss": 0.0946,
+      "step": 21305
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.9688667496886664e-05,
+      "loss": 0.0914,
+      "step": 21306
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.9676214196762136e-05,
+      "loss": 0.1012,
+      "step": 21307
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.96637608966376e-05,
+      "loss": 0.0608,
+      "step": 21308
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.965130759651307e-05,
+      "loss": 0.1375,
+      "step": 21309
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.963885429638854e-05,
+      "loss": 0.2993,
+      "step": 21310
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.962640099626401e-05,
+      "loss": 0.1934,
+      "step": 21311
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.961394769613947e-05,
+      "loss": 0.1464,
+      "step": 21312
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.9601494396014946e-05,
+      "loss": 0.1258,
+      "step": 21313
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.9589041095890405e-05,
+      "loss": 0.1348,
+      "step": 21314
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.957658779576587e-05,
+      "loss": 0.2222,
+      "step": 21315
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.956413449564134e-05,
+      "loss": 0.1533,
+      "step": 21316
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.955168119551681e-05,
+      "loss": 0.1069,
+      "step": 21317
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.9539227895392275e-05,
+      "loss": 0.1181,
+      "step": 21318
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.952677459526774e-05,
+      "loss": 0.1497,
+      "step": 21319
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.9514321295143214e-05,
+      "loss": 0.0952,
+      "step": 21320
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.950186799501868e-05,
+      "loss": 0.1243,
+      "step": 21321
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.948941469489414e-05,
+      "loss": 0.2052,
+      "step": 21322
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.9476961394769605e-05,
+      "loss": 0.1428,
+      "step": 21323
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.946450809464508e-05,
+      "loss": 0.1788,
+      "step": 21324
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.9452054794520544e-05,
+      "loss": 0.0982,
+      "step": 21325
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.943960149439601e-05,
+      "loss": 0.0689,
+      "step": 21326
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.9427148194271476e-05,
+      "loss": 0.1006,
+      "step": 21327
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 5.941469489414695e-05,
+      "loss": 0.1269,
+      "step": 21328
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.940224159402241e-05,
+      "loss": 0.1453,
+      "step": 21329
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.9389788293897873e-05,
+      "loss": 0.1116,
+      "step": 21330
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.937733499377334e-05,
+      "loss": 0.1093,
+      "step": 21331
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.936488169364881e-05,
+      "loss": 0.1148,
+      "step": 21332
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.935242839352428e-05,
+      "loss": 0.1155,
+      "step": 21333
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.9339975093399744e-05,
+      "loss": 0.1184,
+      "step": 21334
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.932752179327522e-05,
+      "loss": 0.2359,
+      "step": 21335
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.931506849315068e-05,
+      "loss": 0.2319,
+      "step": 21336
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.930261519302614e-05,
+      "loss": 0.1228,
+      "step": 21337
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.929016189290161e-05,
+      "loss": 0.1075,
+      "step": 21338
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.927770859277708e-05,
+      "loss": 0.1334,
+      "step": 21339
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.9265255292652546e-05,
+      "loss": 0.1477,
+      "step": 21340
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.925280199252801e-05,
+      "loss": 0.1893,
+      "step": 21341
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.9240348692403485e-05,
+      "loss": 0.1684,
+      "step": 21342
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.922789539227895e-05,
+      "loss": 0.165,
+      "step": 21343
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.921544209215442e-05,
+      "loss": 0.0982,
+      "step": 21344
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.9202988792029876e-05,
+      "loss": 0.0769,
+      "step": 21345
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.919053549190535e-05,
+      "loss": 0.115,
+      "step": 21346
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.9178082191780815e-05,
+      "loss": 0.1222,
+      "step": 21347
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.916562889165628e-05,
+      "loss": 0.1924,
+      "step": 21348
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.9153175591531754e-05,
+      "loss": 0.1508,
+      "step": 21349
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.914072229140722e-05,
+      "loss": 0.1938,
+      "step": 21350
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.9128268991282685e-05,
+      "loss": 0.0579,
+      "step": 21351
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.911581569115816e-05,
+      "loss": 0.145,
+      "step": 21352
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.910336239103362e-05,
+      "loss": 0.2346,
+      "step": 21353
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 5.909090909090908e-05,
+      "loss": 0.1553,
+      "step": 21354
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.907845579078455e-05,
+      "loss": 0.1712,
+      "step": 21355
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.906600249066002e-05,
+      "loss": 0.1381,
+      "step": 21356
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.905354919053549e-05,
+      "loss": 0.1217,
+      "step": 21357
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.9041095890410954e-05,
+      "loss": 0.1204,
+      "step": 21358
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.9028642590286427e-05,
+      "loss": 0.165,
+      "step": 21359
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.901618929016189e-05,
+      "loss": 0.1474,
+      "step": 21360
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.900373599003735e-05,
+      "loss": 0.2072,
+      "step": 21361
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.899128268991282e-05,
+      "loss": 0.083,
+      "step": 21362
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.897882938978829e-05,
+      "loss": 0.1007,
+      "step": 21363
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.8966376089663756e-05,
+      "loss": 0.1065,
+      "step": 21364
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.895392278953922e-05,
+      "loss": 0.1909,
+      "step": 21365
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.8941469489414695e-05,
+      "loss": 0.2035,
+      "step": 21366
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.892901618929016e-05,
+      "loss": 0.1311,
+      "step": 21367
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.891656288916563e-05,
+      "loss": 0.1886,
+      "step": 21368
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.8904109589041086e-05,
+      "loss": 0.0978,
+      "step": 21369
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.889165628891656e-05,
+      "loss": 0.1012,
+      "step": 21370
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.8879202988792025e-05,
+      "loss": 0.1389,
+      "step": 21371
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.886674968866749e-05,
+      "loss": 0.2165,
+      "step": 21372
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.8854296388542957e-05,
+      "loss": 0.1561,
+      "step": 21373
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.884184308841843e-05,
+      "loss": 0.1329,
+      "step": 21374
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.8829389788293895e-05,
+      "loss": 0.0985,
+      "step": 21375
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.881693648816936e-05,
+      "loss": 0.0887,
+      "step": 21376
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.880448318804482e-05,
+      "loss": 0.095,
+      "step": 21377
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.879202988792029e-05,
+      "loss": 0.1822,
+      "step": 21378
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.877957658779576e-05,
+      "loss": 0.1897,
+      "step": 21379
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 5.8767123287671225e-05,
+      "loss": 0.1557,
+      "step": 21380
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.87546699875467e-05,
+      "loss": 0.1334,
+      "step": 21381
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.8742216687422164e-05,
+      "loss": 0.1256,
+      "step": 21382
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.872976338729763e-05,
+      "loss": 0.1019,
+      "step": 21383
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.871731008717309e-05,
+      "loss": 0.1376,
+      "step": 21384
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.870485678704856e-05,
+      "loss": 0.1626,
+      "step": 21385
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.869240348692403e-05,
+      "loss": 0.1564,
+      "step": 21386
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.867995018679949e-05,
+      "loss": 0.1857,
+      "step": 21387
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.8667496886674966e-05,
+      "loss": 0.1131,
+      "step": 21388
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.865504358655043e-05,
+      "loss": 0.1277,
+      "step": 21389
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.86425902864259e-05,
+      "loss": 0.0993,
+      "step": 21390
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.863013698630137e-05,
+      "loss": 0.111,
+      "step": 21391
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.861768368617683e-05,
+      "loss": 0.1779,
+      "step": 21392
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.8605230386052296e-05,
+      "loss": 0.1058,
+      "step": 21393
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.859277708592776e-05,
+      "loss": 0.1621,
+      "step": 21394
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.8580323785803234e-05,
+      "loss": 0.1335,
+      "step": 21395
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.85678704856787e-05,
+      "loss": 0.0909,
+      "step": 21396
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.8555417185554166e-05,
+      "loss": 0.1595,
+      "step": 21397
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.854296388542964e-05,
+      "loss": 0.1385,
+      "step": 21398
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.8530510585305105e-05,
+      "loss": 0.241,
+      "step": 21399
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.8518057285180564e-05,
+      "loss": 0.1026,
+      "step": 21400
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.850560398505603e-05,
+      "loss": 0.1078,
+      "step": 21401
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.84931506849315e-05,
+      "loss": 0.1926,
+      "step": 21402
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.848069738480697e-05,
+      "loss": 0.1092,
+      "step": 21403
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.8468244084682435e-05,
+      "loss": 0.1818,
+      "step": 21404
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.845579078455791e-05,
+      "loss": 0.152,
+      "step": 21405
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 5.8443337484433373e-05,
+      "loss": 0.1533,
+      "step": 21406
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.843088418430884e-05,
+      "loss": 0.112,
+      "step": 21407
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.84184308841843e-05,
+      "loss": 0.1385,
+      "step": 21408
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.840597758405977e-05,
+      "loss": 0.1155,
+      "step": 21409
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.839352428393524e-05,
+      "loss": 0.2626,
+      "step": 21410
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.83810709838107e-05,
+      "loss": 0.0973,
+      "step": 21411
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.836861768368617e-05,
+      "loss": 0.118,
+      "step": 21412
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.835616438356164e-05,
+      "loss": 0.1089,
+      "step": 21413
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.834371108343711e-05,
+      "loss": 0.0779,
+      "step": 21414
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.8331257783312574e-05,
+      "loss": 0.0943,
+      "step": 21415
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.831880448318804e-05,
+      "loss": 0.0742,
+      "step": 21416
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.8306351183063506e-05,
+      "loss": 0.1594,
+      "step": 21417
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.829389788293897e-05,
+      "loss": 0.115,
+      "step": 21418
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.828144458281444e-05,
+      "loss": 0.1084,
+      "step": 21419
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.826899128268991e-05,
+      "loss": 0.1437,
+      "step": 21420
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.8256537982565376e-05,
+      "loss": 0.0901,
+      "step": 21421
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.824408468244084e-05,
+      "loss": 0.1807,
+      "step": 21422
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.8231631382316315e-05,
+      "loss": 0.1607,
+      "step": 21423
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.8219178082191774e-05,
+      "loss": 0.2466,
+      "step": 21424
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.820672478206724e-05,
+      "loss": 0.1493,
+      "step": 21425
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.8194271481942706e-05,
+      "loss": 0.2073,
+      "step": 21426
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.818181818181818e-05,
+      "loss": 0.1085,
+      "step": 21427
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.8169364881693645e-05,
+      "loss": 0.2202,
+      "step": 21428
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.815691158156911e-05,
+      "loss": 0.1375,
+      "step": 21429
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.814445828144458e-05,
+      "loss": 0.1632,
+      "step": 21430
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.813200498132004e-05,
+      "loss": 0.1263,
+      "step": 21431
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 5.811955168119551e-05,
+      "loss": 0.1063,
+      "step": 21432
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.8107098381070974e-05,
+      "loss": 0.0981,
+      "step": 21433
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.809464508094645e-05,
+      "loss": 0.0914,
+      "step": 21434
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.808219178082191e-05,
+      "loss": 0.2179,
+      "step": 21435
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.806973848069738e-05,
+      "loss": 0.1368,
+      "step": 21436
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.805728518057285e-05,
+      "loss": 0.126,
+      "step": 21437
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.804483188044832e-05,
+      "loss": 0.0765,
+      "step": 21438
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.803237858032378e-05,
+      "loss": 0.0898,
+      "step": 21439
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.801992528019924e-05,
+      "loss": 0.0865,
+      "step": 21440
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.8007471980074715e-05,
+      "loss": 0.1855,
+      "step": 21441
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.799501867995018e-05,
+      "loss": 0.1399,
+      "step": 21442
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.798256537982565e-05,
+      "loss": 0.1458,
+      "step": 21443
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.797011207970112e-05,
+      "loss": 0.1294,
+      "step": 21444
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.7957658779576586e-05,
+      "loss": 0.1624,
+      "step": 21445
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.794520547945205e-05,
+      "loss": 0.0697,
+      "step": 21446
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.793275217932751e-05,
+      "loss": 0.1493,
+      "step": 21447
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.7920298879202984e-05,
+      "loss": 0.2054,
+      "step": 21448
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.790784557907845e-05,
+      "loss": 0.1269,
+      "step": 21449
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.7895392278953916e-05,
+      "loss": 0.1758,
+      "step": 21450
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.788293897882939e-05,
+      "loss": 0.0968,
+      "step": 21451
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.7870485678704854e-05,
+      "loss": 0.1695,
+      "step": 21452
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.785803237858032e-05,
+      "loss": 0.1388,
+      "step": 21453
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.7845579078455786e-05,
+      "loss": 0.2188,
+      "step": 21454
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.783312577833125e-05,
+      "loss": 0.255,
+      "step": 21455
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.782067247820672e-05,
+      "loss": 0.1552,
+      "step": 21456
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.7808219178082184e-05,
+      "loss": 0.1275,
+      "step": 21457
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.779576587795765e-05,
+      "loss": 0.0855,
+      "step": 21458
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 5.778331257783312e-05,
+      "loss": 0.1376,
+      "step": 21459
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.777085927770859e-05,
+      "loss": 0.1828,
+      "step": 21460
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.7758405977584055e-05,
+      "loss": 0.1443,
+      "step": 21461
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.774595267745953e-05,
+      "loss": 0.1217,
+      "step": 21462
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.7733499377334986e-05,
+      "loss": 0.1661,
+      "step": 21463
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.772104607721045e-05,
+      "loss": 0.0958,
+      "step": 21464
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.770859277708592e-05,
+      "loss": 0.058,
+      "step": 21465
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.769613947696139e-05,
+      "loss": 0.1817,
+      "step": 21466
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.768368617683686e-05,
+      "loss": 0.144,
+      "step": 21467
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.767123287671232e-05,
+      "loss": 0.1144,
+      "step": 21468
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.7658779576587796e-05,
+      "loss": 0.1707,
+      "step": 21469
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.764632627646326e-05,
+      "loss": 0.0881,
+      "step": 21470
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.763387297633872e-05,
+      "loss": 0.1267,
+      "step": 21471
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.762141967621419e-05,
+      "loss": 0.1222,
+      "step": 21472
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.760896637608966e-05,
+      "loss": 0.1433,
+      "step": 21473
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.7596513075965125e-05,
+      "loss": 0.1419,
+      "step": 21474
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.758405977584059e-05,
+      "loss": 0.1982,
+      "step": 21475
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.7571606475716064e-05,
+      "loss": 0.0688,
+      "step": 21476
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.755915317559153e-05,
+      "loss": 0.0692,
+      "step": 21477
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.7546699875466996e-05,
+      "loss": 0.1771,
+      "step": 21478
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.7534246575342455e-05,
+      "loss": 0.2116,
+      "step": 21479
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.752179327521793e-05,
+      "loss": 0.1039,
+      "step": 21480
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.7509339975093394e-05,
+      "loss": 0.1979,
+      "step": 21481
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.749688667496886e-05,
+      "loss": 0.0797,
+      "step": 21482
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.748443337484433e-05,
+      "loss": 0.0918,
+      "step": 21483
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.74719800747198e-05,
+      "loss": 0.1002,
+      "step": 21484
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 5.7459526774595264e-05,
+      "loss": 0.299,
+      "step": 21485
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.7447073474470724e-05,
+      "loss": 0.1629,
+      "step": 21486
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.7434620174346196e-05,
+      "loss": 0.091,
+      "step": 21487
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.742216687422166e-05,
+      "loss": 0.0971,
+      "step": 21488
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.740971357409713e-05,
+      "loss": 0.1181,
+      "step": 21489
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.73972602739726e-05,
+      "loss": 0.0672,
+      "step": 21490
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.738480697384807e-05,
+      "loss": 0.1646,
+      "step": 21491
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.737235367372353e-05,
+      "loss": 0.1083,
+      "step": 21492
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.7359900373599e-05,
+      "loss": 0.1477,
+      "step": 21493
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.7347447073474465e-05,
+      "loss": 0.1771,
+      "step": 21494
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.733499377334993e-05,
+      "loss": 0.0656,
+      "step": 21495
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.7322540473225397e-05,
+      "loss": 0.1429,
+      "step": 21496
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.731008717310087e-05,
+      "loss": 0.1479,
+      "step": 21497
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.7297633872976335e-05,
+      "loss": 0.2263,
+      "step": 21498
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.72851805728518e-05,
+      "loss": 0.0996,
+      "step": 21499
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.727272727272727e-05,
+      "loss": 0.1207,
+      "step": 21500
+    },
+    {
+      "epoch": 8.24,
+      "eval_cer": 0.10079013598142693,
+      "eval_loss": 0.4120646119117737,
+      "eval_runtime": 47.7401,
+      "eval_samples_per_second": 36.72,
+      "eval_steps_per_second": 4.608,
+      "eval_wer": 0.16056478254966738,
+      "step": 21500
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.726027397260274e-05,
+      "loss": 0.1602,
+      "step": 21501
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.72478206724782e-05,
+      "loss": 0.0989,
+      "step": 21502
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.7235367372353665e-05,
+      "loss": 0.1025,
+      "step": 21503
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.722291407222913e-05,
+      "loss": 0.227,
+      "step": 21504
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.7210460772104604e-05,
+      "loss": 0.2282,
+      "step": 21505
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.719800747198007e-05,
+      "loss": 0.1256,
+      "step": 21506
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.7185554171855535e-05,
+      "loss": 0.0854,
+      "step": 21507
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.717310087173101e-05,
+      "loss": 0.1916,
+      "step": 21508
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.7160647571606474e-05,
+      "loss": 0.161,
+      "step": 21509
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.714819427148193e-05,
+      "loss": 0.2127,
+      "step": 21510
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 5.71357409713574e-05,
+      "loss": 0.169,
+      "step": 21511
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.712328767123287e-05,
+      "loss": 0.1943,
+      "step": 21512
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.711083437110834e-05,
+      "loss": 0.14,
+      "step": 21513
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.7098381070983804e-05,
+      "loss": 0.0874,
+      "step": 21514
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.7085927770859277e-05,
+      "loss": 0.0917,
+      "step": 21515
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.707347447073474e-05,
+      "loss": 0.133,
+      "step": 21516
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.706102117061021e-05,
+      "loss": 0.1239,
+      "step": 21517
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.704856787048567e-05,
+      "loss": 0.1327,
+      "step": 21518
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.703611457036114e-05,
+      "loss": 0.1502,
+      "step": 21519
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.7023661270236606e-05,
+      "loss": 0.2083,
+      "step": 21520
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.701120797011207e-05,
+      "loss": 0.1152,
+      "step": 21521
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.6998754669987545e-05,
+      "loss": 0.1897,
+      "step": 21522
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.698630136986301e-05,
+      "loss": 0.1663,
+      "step": 21523
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.697384806973848e-05,
+      "loss": 0.2003,
+      "step": 21524
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.696139476961395e-05,
+      "loss": 0.1791,
+      "step": 21525
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.694894146948941e-05,
+      "loss": 0.1298,
+      "step": 21526
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.6936488169364875e-05,
+      "loss": 0.1079,
+      "step": 21527
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.692403486924034e-05,
+      "loss": 0.1242,
+      "step": 21528
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.691158156911581e-05,
+      "loss": 0.2025,
+      "step": 21529
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.689912826899128e-05,
+      "loss": 0.1211,
+      "step": 21530
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.6886674968866745e-05,
+      "loss": 0.1649,
+      "step": 21531
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.687422166874222e-05,
+      "loss": 0.1294,
+      "step": 21532
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.6861768368617684e-05,
+      "loss": 0.1322,
+      "step": 21533
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.684931506849314e-05,
+      "loss": 0.0853,
+      "step": 21534
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.683686176836861e-05,
+      "loss": 0.2549,
+      "step": 21535
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.682440846824408e-05,
+      "loss": 0.2114,
+      "step": 21536
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 5.681195516811955e-05,
+      "loss": 0.1444,
+      "step": 21537
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.6799501867995014e-05,
+      "loss": 0.1069,
+      "step": 21538
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.678704856787048e-05,
+      "loss": 0.0688,
+      "step": 21539
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.677459526774595e-05,
+      "loss": 0.1235,
+      "step": 21540
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.676214196762141e-05,
+      "loss": 0.1503,
+      "step": 21541
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.674968866749688e-05,
+      "loss": 0.1226,
+      "step": 21542
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.673723536737234e-05,
+      "loss": 0.1247,
+      "step": 21543
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.6724782067247816e-05,
+      "loss": 0.1519,
+      "step": 21544
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.671232876712328e-05,
+      "loss": 0.1054,
+      "step": 21545
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.669987546699875e-05,
+      "loss": 0.0766,
+      "step": 21546
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.668742216687422e-05,
+      "loss": 0.1075,
+      "step": 21547
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.667496886674969e-05,
+      "loss": 0.2041,
+      "step": 21548
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.6662515566625146e-05,
+      "loss": 0.1162,
+      "step": 21549
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.665006226650061e-05,
+      "loss": 0.1184,
+      "step": 21550
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.6637608966376084e-05,
+      "loss": 0.1131,
+      "step": 21551
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.662515566625155e-05,
+      "loss": 0.1403,
+      "step": 21552
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.6612702366127016e-05,
+      "loss": 0.1383,
+      "step": 21553
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.660024906600249e-05,
+      "loss": 0.1765,
+      "step": 21554
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.6587795765877955e-05,
+      "loss": 0.1426,
+      "step": 21555
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.657534246575342e-05,
+      "loss": 0.0786,
+      "step": 21556
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.656288916562888e-05,
+      "loss": 0.1172,
+      "step": 21557
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.655043586550435e-05,
+      "loss": 0.0966,
+      "step": 21558
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.653798256537982e-05,
+      "loss": 0.1681,
+      "step": 21559
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.6525529265255285e-05,
+      "loss": 0.1712,
+      "step": 21560
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.651307596513076e-05,
+      "loss": 0.1558,
+      "step": 21561
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.6500622665006223e-05,
+      "loss": 0.1493,
+      "step": 21562
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 5.648816936488169e-05,
+      "loss": 0.0916,
+      "step": 21563
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.647571606475716e-05,
+      "loss": 0.1202,
+      "step": 21564
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.646326276463262e-05,
+      "loss": 0.1471,
+      "step": 21565
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.645080946450809e-05,
+      "loss": 0.1783,
+      "step": 21566
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.643835616438355e-05,
+      "loss": 0.3194,
+      "step": 21567
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.6425902864259026e-05,
+      "loss": 0.2781,
+      "step": 21568
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.641344956413449e-05,
+      "loss": 0.0813,
+      "step": 21569
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.640099626400996e-05,
+      "loss": 0.1155,
+      "step": 21570
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.638854296388543e-05,
+      "loss": 0.0868,
+      "step": 21571
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.6376089663760896e-05,
+      "loss": 0.1474,
+      "step": 21572
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.6363636363636356e-05,
+      "loss": 0.1391,
+      "step": 21573
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.635118306351182e-05,
+      "loss": 0.1942,
+      "step": 21574
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.6338729763387294e-05,
+      "loss": 0.0784,
+      "step": 21575
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.632627646326276e-05,
+      "loss": 0.0961,
+      "step": 21576
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.6313823163138226e-05,
+      "loss": 0.0461,
+      "step": 21577
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.63013698630137e-05,
+      "loss": 0.0865,
+      "step": 21578
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.6288916562889165e-05,
+      "loss": 0.2534,
+      "step": 21579
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.627646326276463e-05,
+      "loss": 0.144,
+      "step": 21580
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.626400996264009e-05,
+      "loss": 0.1721,
+      "step": 21581
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.625155666251556e-05,
+      "loss": 0.0724,
+      "step": 21582
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.623910336239103e-05,
+      "loss": 0.075,
+      "step": 21583
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.6226650062266495e-05,
+      "loss": 0.0979,
+      "step": 21584
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.621419676214196e-05,
+      "loss": 0.142,
+      "step": 21585
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.620174346201743e-05,
+      "loss": 0.1228,
+      "step": 21586
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.61892901618929e-05,
+      "loss": 0.116,
+      "step": 21587
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.6176836861768365e-05,
+      "loss": 0.1206,
+      "step": 21588
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 5.6164383561643824e-05,
+      "loss": 0.0886,
+      "step": 21589
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.61519302615193e-05,
+      "loss": 0.0889,
+      "step": 21590
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.613947696139476e-05,
+      "loss": 0.1825,
+      "step": 21591
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.612702366127023e-05,
+      "loss": 0.2089,
+      "step": 21592
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.61145703611457e-05,
+      "loss": 0.1024,
+      "step": 21593
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.610211706102117e-05,
+      "loss": 0.1841,
+      "step": 21594
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.6089663760896634e-05,
+      "loss": 0.1666,
+      "step": 21595
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.607721046077209e-05,
+      "loss": 0.0996,
+      "step": 21596
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.6064757160647565e-05,
+      "loss": 0.2644,
+      "step": 21597
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.605230386052303e-05,
+      "loss": 0.2442,
+      "step": 21598
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.60398505603985e-05,
+      "loss": 0.1034,
+      "step": 21599
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.602739726027397e-05,
+      "loss": 0.2023,
+      "step": 21600
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.6014943960149436e-05,
+      "loss": 0.0805,
+      "step": 21601
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.60024906600249e-05,
+      "loss": 0.1395,
+      "step": 21602
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.5990037359900375e-05,
+      "loss": 0.0861,
+      "step": 21603
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.5977584059775834e-05,
+      "loss": 0.1734,
+      "step": 21604
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.59651307596513e-05,
+      "loss": 0.1685,
+      "step": 21605
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.5952677459526766e-05,
+      "loss": 0.1129,
+      "step": 21606
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.594022415940224e-05,
+      "loss": 0.0806,
+      "step": 21607
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.5927770859277704e-05,
+      "loss": 0.1952,
+      "step": 21608
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.591531755915317e-05,
+      "loss": 0.2284,
+      "step": 21609
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.590286425902864e-05,
+      "loss": 0.2364,
+      "step": 21610
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.589041095890411e-05,
+      "loss": 0.1274,
+      "step": 21611
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.587795765877957e-05,
+      "loss": 0.0973,
+      "step": 21612
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.5865504358655034e-05,
+      "loss": 0.194,
+      "step": 21613
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.585305105853051e-05,
+      "loss": 0.1758,
+      "step": 21614
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 5.584059775840597e-05,
+      "loss": 0.1523,
+      "step": 21615
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.582814445828144e-05,
+      "loss": 0.2,
+      "step": 21616
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.581569115815691e-05,
+      "loss": 0.0943,
+      "step": 21617
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.580323785803238e-05,
+      "loss": 0.1559,
+      "step": 21618
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.579078455790784e-05,
+      "loss": 0.1737,
+      "step": 21619
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.57783312577833e-05,
+      "loss": 0.1785,
+      "step": 21620
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.5765877957658775e-05,
+      "loss": 0.1784,
+      "step": 21621
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.575342465753424e-05,
+      "loss": 0.1544,
+      "step": 21622
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.574097135740971e-05,
+      "loss": 0.1656,
+      "step": 21623
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.572851805728517e-05,
+      "loss": 0.1536,
+      "step": 21624
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.5716064757160646e-05,
+      "loss": 0.1224,
+      "step": 21625
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.570361145703611e-05,
+      "loss": 0.1024,
+      "step": 21626
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.569115815691158e-05,
+      "loss": 0.1498,
+      "step": 21627
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.5678704856787044e-05,
+      "loss": 0.2113,
+      "step": 21628
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.566625155666251e-05,
+      "loss": 0.3077,
+      "step": 21629
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.5653798256537975e-05,
+      "loss": 0.143,
+      "step": 21630
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.564134495641344e-05,
+      "loss": 0.1105,
+      "step": 21631
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.5628891656288914e-05,
+      "loss": 0.0899,
+      "step": 21632
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.561643835616438e-05,
+      "loss": 0.0974,
+      "step": 21633
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.5603985056039846e-05,
+      "loss": 0.0974,
+      "step": 21634
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.559153175591532e-05,
+      "loss": 0.2579,
+      "step": 21635
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.557907845579078e-05,
+      "loss": 0.1117,
+      "step": 21636
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.5566625155666244e-05,
+      "loss": 0.1798,
+      "step": 21637
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.555417185554171e-05,
+      "loss": 0.2687,
+      "step": 21638
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.554171855541718e-05,
+      "loss": 0.0811,
+      "step": 21639
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.552926525529265e-05,
+      "loss": 0.1623,
+      "step": 21640
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 5.5516811955168114e-05,
+      "loss": 0.1485,
+      "step": 21641
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.550435865504359e-05,
+      "loss": 0.1214,
+      "step": 21642
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.5491905354919046e-05,
+      "loss": 0.1823,
+      "step": 21643
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.547945205479451e-05,
+      "loss": 0.1234,
+      "step": 21644
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.546699875466998e-05,
+      "loss": 0.0948,
+      "step": 21645
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.545454545454545e-05,
+      "loss": 0.1041,
+      "step": 21646
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.544209215442092e-05,
+      "loss": 0.1272,
+      "step": 21647
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.542963885429638e-05,
+      "loss": 0.1574,
+      "step": 21648
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.5417185554171856e-05,
+      "loss": 0.1087,
+      "step": 21649
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.540473225404732e-05,
+      "loss": 0.188,
+      "step": 21650
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.539227895392278e-05,
+      "loss": 0.1553,
+      "step": 21651
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.5379825653798247e-05,
+      "loss": 0.0737,
+      "step": 21652
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.536737235367372e-05,
+      "loss": 0.1004,
+      "step": 21653
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.5354919053549185e-05,
+      "loss": 0.1802,
+      "step": 21654
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.534246575342465e-05,
+      "loss": 0.158,
+      "step": 21655
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.5330012453300124e-05,
+      "loss": 0.1426,
+      "step": 21656
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.531755915317559e-05,
+      "loss": 0.1997,
+      "step": 21657
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.5305105853051056e-05,
+      "loss": 0.1134,
+      "step": 21658
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.5292652552926515e-05,
+      "loss": 0.0653,
+      "step": 21659
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.528019925280199e-05,
+      "loss": 0.1329,
+      "step": 21660
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.5267745952677454e-05,
+      "loss": 0.1269,
+      "step": 21661
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.525529265255292e-05,
+      "loss": 0.1051,
+      "step": 21662
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.524283935242839e-05,
+      "loss": 0.0717,
+      "step": 21663
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.523038605230386e-05,
+      "loss": 0.0703,
+      "step": 21664
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.5217932752179324e-05,
+      "loss": 0.0985,
+      "step": 21665
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.520547945205479e-05,
+      "loss": 0.2117,
+      "step": 21666
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 5.5193026151930256e-05,
+      "loss": 0.1209,
+      "step": 21667
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.518057285180572e-05,
+      "loss": 0.0983,
+      "step": 21668
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.516811955168119e-05,
+      "loss": 0.0933,
+      "step": 21669
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.5155666251556654e-05,
+      "loss": 0.1003,
+      "step": 21670
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.514321295143213e-05,
+      "loss": 0.0654,
+      "step": 21671
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.513075965130759e-05,
+      "loss": 0.1797,
+      "step": 21672
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.511830635118306e-05,
+      "loss": 0.1834,
+      "step": 21673
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.510585305105853e-05,
+      "loss": 0.1424,
+      "step": 21674
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.509339975093399e-05,
+      "loss": 0.1591,
+      "step": 21675
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.5080946450809456e-05,
+      "loss": 0.1297,
+      "step": 21676
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.506849315068492e-05,
+      "loss": 0.1023,
+      "step": 21677
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.5056039850560395e-05,
+      "loss": 0.1094,
+      "step": 21678
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.504358655043586e-05,
+      "loss": 0.1619,
+      "step": 21679
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.503113325031133e-05,
+      "loss": 0.2025,
+      "step": 21680
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.50186799501868e-05,
+      "loss": 0.1052,
+      "step": 21681
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.5006226650062266e-05,
+      "loss": 0.1347,
+      "step": 21682
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.4993773349937725e-05,
+      "loss": 0.0677,
+      "step": 21683
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.498132004981319e-05,
+      "loss": 0.1195,
+      "step": 21684
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.4968866749688663e-05,
+      "loss": 0.2146,
+      "step": 21685
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.495641344956413e-05,
+      "loss": 0.088,
+      "step": 21686
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.4943960149439595e-05,
+      "loss": 0.0876,
+      "step": 21687
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.493150684931507e-05,
+      "loss": 0.078,
+      "step": 21688
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.4919053549190534e-05,
+      "loss": 0.0851,
+      "step": 21689
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.4906600249066e-05,
+      "loss": 0.0967,
+      "step": 21690
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.489414694894146e-05,
+      "loss": 0.2099,
+      "step": 21691
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.488169364881693e-05,
+      "loss": 0.2306,
+      "step": 21692
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 5.48692403486924e-05,
+      "loss": 0.1001,
+      "step": 21693
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4856787048567864e-05,
+      "loss": 0.1201,
+      "step": 21694
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4844333748443336e-05,
+      "loss": 0.1153,
+      "step": 21695
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.48318804483188e-05,
+      "loss": 0.1003,
+      "step": 21696
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.481942714819427e-05,
+      "loss": 0.1801,
+      "step": 21697
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.480697384806973e-05,
+      "loss": 0.1589,
+      "step": 21698
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.47945205479452e-05,
+      "loss": 0.0971,
+      "step": 21699
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4782067247820666e-05,
+      "loss": 0.1244,
+      "step": 21700
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.476961394769613e-05,
+      "loss": 0.0719,
+      "step": 21701
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4757160647571605e-05,
+      "loss": 0.081,
+      "step": 21702
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.474470734744707e-05,
+      "loss": 0.1471,
+      "step": 21703
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.473225404732254e-05,
+      "loss": 0.2155,
+      "step": 21704
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4719800747198e-05,
+      "loss": 0.1386,
+      "step": 21705
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.470734744707347e-05,
+      "loss": 0.0724,
+      "step": 21706
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4694894146948935e-05,
+      "loss": 0.0857,
+      "step": 21707
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.46824408468244e-05,
+      "loss": 0.0729,
+      "step": 21708
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.466998754669987e-05,
+      "loss": 0.0845,
+      "step": 21709
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.465753424657534e-05,
+      "loss": 0.2033,
+      "step": 21710
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4645080946450805e-05,
+      "loss": 0.1414,
+      "step": 21711
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.463262764632627e-05,
+      "loss": 0.1592,
+      "step": 21712
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4620174346201744e-05,
+      "loss": 0.1067,
+      "step": 21713
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.46077210460772e-05,
+      "loss": 0.1117,
+      "step": 21714
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.459526774595267e-05,
+      "loss": 0.1628,
+      "step": 21715
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4582814445828135e-05,
+      "loss": 0.2337,
+      "step": 21716
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.457036114570361e-05,
+      "loss": 0.1803,
+      "step": 21717
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.4557907845579073e-05,
+      "loss": 0.1719,
+      "step": 21718
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 5.454545454545454e-05,
+      "loss": 0.1394,
+      "step": 21719
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.453300124533001e-05,
+      "loss": 0.1307,
+      "step": 21720
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.452054794520548e-05,
+      "loss": 0.058,
+      "step": 21721
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.450809464508094e-05,
+      "loss": 0.0943,
+      "step": 21722
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.44956413449564e-05,
+      "loss": 0.1642,
+      "step": 21723
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4483188044831876e-05,
+      "loss": 0.1212,
+      "step": 21724
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.447073474470734e-05,
+      "loss": 0.1031,
+      "step": 21725
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.445828144458281e-05,
+      "loss": 0.0692,
+      "step": 21726
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.444582814445828e-05,
+      "loss": 0.0872,
+      "step": 21727
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4433374844333747e-05,
+      "loss": 0.0897,
+      "step": 21728
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.442092154420921e-05,
+      "loss": 0.2022,
+      "step": 21729
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.440846824408467e-05,
+      "loss": 0.1454,
+      "step": 21730
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4396014943960144e-05,
+      "loss": 0.1994,
+      "step": 21731
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.438356164383561e-05,
+      "loss": 0.1356,
+      "step": 21732
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4371108343711076e-05,
+      "loss": 0.1018,
+      "step": 21733
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.435865504358655e-05,
+      "loss": 0.1562,
+      "step": 21734
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4346201743462015e-05,
+      "loss": 0.2818,
+      "step": 21735
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.433374844333748e-05,
+      "loss": 0.1204,
+      "step": 21736
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4321295143212954e-05,
+      "loss": 0.0847,
+      "step": 21737
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.430884184308841e-05,
+      "loss": 0.1486,
+      "step": 21738
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.429638854296388e-05,
+      "loss": 0.1639,
+      "step": 21739
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.4283935242839345e-05,
+      "loss": 0.2115,
+      "step": 21740
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.427148194271482e-05,
+      "loss": 0.2324,
+      "step": 21741
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.425902864259028e-05,
+      "loss": 0.1899,
+      "step": 21742
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.424657534246575e-05,
+      "loss": 0.1673,
+      "step": 21743
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.423412204234122e-05,
+      "loss": 0.1066,
+      "step": 21744
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.422166874221669e-05,
+      "loss": 0.2101,
+      "step": 21745
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 5.420921544209215e-05,
+      "loss": 0.1015,
+      "step": 21746
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.419676214196761e-05,
+      "loss": 0.1492,
+      "step": 21747
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4184308841843086e-05,
+      "loss": 0.175,
+      "step": 21748
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.417185554171855e-05,
+      "loss": 0.1058,
+      "step": 21749
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.415940224159402e-05,
+      "loss": 0.2121,
+      "step": 21750
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4146948941469484e-05,
+      "loss": 0.1245,
+      "step": 21751
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4134495641344956e-05,
+      "loss": 0.07,
+      "step": 21752
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4122042341220415e-05,
+      "loss": 0.1749,
+      "step": 21753
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.410958904109588e-05,
+      "loss": 0.2061,
+      "step": 21754
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.409713574097135e-05,
+      "loss": 0.2557,
+      "step": 21755
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.408468244084682e-05,
+      "loss": 0.1125,
+      "step": 21756
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4072229140722286e-05,
+      "loss": 0.0779,
+      "step": 21757
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.405977584059775e-05,
+      "loss": 0.1085,
+      "step": 21758
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4047322540473225e-05,
+      "loss": 0.0746,
+      "step": 21759
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.403486924034869e-05,
+      "loss": 0.1647,
+      "step": 21760
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.402241594022415e-05,
+      "loss": 0.2009,
+      "step": 21761
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.4009962640099616e-05,
+      "loss": 0.1391,
+      "step": 21762
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.399750933997509e-05,
+      "loss": 0.1167,
+      "step": 21763
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.3985056039850554e-05,
+      "loss": 0.1945,
+      "step": 21764
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.397260273972602e-05,
+      "loss": 0.1182,
+      "step": 21765
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.396014943960149e-05,
+      "loss": 0.2362,
+      "step": 21766
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.394769613947696e-05,
+      "loss": 0.1491,
+      "step": 21767
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.3935242839352425e-05,
+      "loss": 0.1086,
+      "step": 21768
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.3922789539227884e-05,
+      "loss": 0.1121,
+      "step": 21769
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.391033623910336e-05,
+      "loss": 0.1256,
+      "step": 21770
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.389788293897882e-05,
+      "loss": 0.0799,
+      "step": 21771
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 5.388542963885429e-05,
+      "loss": 0.1713,
+      "step": 21772
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.387297633872976e-05,
+      "loss": 0.1278,
+      "step": 21773
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.386052303860523e-05,
+      "loss": 0.1161,
+      "step": 21774
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.384806973848069e-05,
+      "loss": 0.0793,
+      "step": 21775
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.3835616438356166e-05,
+      "loss": 0.1901,
+      "step": 21776
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.3823163138231625e-05,
+      "loss": 0.1069,
+      "step": 21777
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.381070983810709e-05,
+      "loss": 0.0763,
+      "step": 21778
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.379825653798256e-05,
+      "loss": 0.2714,
+      "step": 21779
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.378580323785803e-05,
+      "loss": 0.2165,
+      "step": 21780
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.3773349937733496e-05,
+      "loss": 0.1216,
+      "step": 21781
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.376089663760896e-05,
+      "loss": 0.0868,
+      "step": 21782
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.3748443337484434e-05,
+      "loss": 0.1612,
+      "step": 21783
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.37359900373599e-05,
+      "loss": 0.1547,
+      "step": 21784
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.37359900373599e-05,
+      "loss": 0.8895,
+      "step": 21785
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.372353673723536e-05,
+      "loss": 0.1923,
+      "step": 21786
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.3711083437110826e-05,
+      "loss": 0.1254,
+      "step": 21787
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.36986301369863e-05,
+      "loss": 0.1196,
+      "step": 21788
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.3686176836861764e-05,
+      "loss": 0.1227,
+      "step": 21789
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.367372353673723e-05,
+      "loss": 0.0858,
+      "step": 21790
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.36612702366127e-05,
+      "loss": 0.1667,
+      "step": 21791
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.364881693648817e-05,
+      "loss": 0.1377,
+      "step": 21792
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.3636363636363635e-05,
+      "loss": 0.2879,
+      "step": 21793
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.3623910336239094e-05,
+      "loss": 0.197,
+      "step": 21794
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.361145703611457e-05,
+      "loss": 0.0953,
+      "step": 21795
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.359900373599003e-05,
+      "loss": 0.1279,
+      "step": 21796
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.35865504358655e-05,
+      "loss": 0.1908,
+      "step": 21797
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 5.3574097135740964e-05,
+      "loss": 0.1927,
+      "step": 21798
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.356164383561644e-05,
+      "loss": 0.1176,
+      "step": 21799
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.35491905354919e-05,
+      "loss": 0.1122,
+      "step": 21800
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.353673723536737e-05,
+      "loss": 0.1806,
+      "step": 21801
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.352428393524283e-05,
+      "loss": 0.1001,
+      "step": 21802
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.35118306351183e-05,
+      "loss": 0.131,
+      "step": 21803
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.349937733499377e-05,
+      "loss": 0.198,
+      "step": 21804
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.348692403486923e-05,
+      "loss": 0.131,
+      "step": 21805
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.3474470734744706e-05,
+      "loss": 0.1349,
+      "step": 21806
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.346201743462017e-05,
+      "loss": 0.1114,
+      "step": 21807
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.344956413449564e-05,
+      "loss": 0.1403,
+      "step": 21808
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.34371108343711e-05,
+      "loss": 0.1078,
+      "step": 21809
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.342465753424657e-05,
+      "loss": 0.1899,
+      "step": 21810
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.3412204234122035e-05,
+      "loss": 0.1865,
+      "step": 21811
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.33997509339975e-05,
+      "loss": 0.1217,
+      "step": 21812
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.3387297633872974e-05,
+      "loss": 0.1328,
+      "step": 21813
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.337484433374844e-05,
+      "loss": 0.0709,
+      "step": 21814
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.3362391033623906e-05,
+      "loss": 0.1187,
+      "step": 21815
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.334993773349938e-05,
+      "loss": 0.1856,
+      "step": 21816
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.333748443337484e-05,
+      "loss": 0.2255,
+      "step": 21817
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.3325031133250304e-05,
+      "loss": 0.2162,
+      "step": 21818
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.331257783312577e-05,
+      "loss": 0.1433,
+      "step": 21819
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.330012453300124e-05,
+      "loss": 0.181,
+      "step": 21820
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.328767123287671e-05,
+      "loss": 0.0767,
+      "step": 21821
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.3275217932752174e-05,
+      "loss": 0.1934,
+      "step": 21822
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.326276463262765e-05,
+      "loss": 0.2527,
+      "step": 21823
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 5.325031133250311e-05,
+      "loss": 0.1384,
+      "step": 21824
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.323785803237857e-05,
+      "loss": 0.1283,
+      "step": 21825
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.322540473225404e-05,
+      "loss": 0.1763,
+      "step": 21826
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.321295143212951e-05,
+      "loss": 0.1375,
+      "step": 21827
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.320049813200498e-05,
+      "loss": 0.1005,
+      "step": 21828
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.318804483188044e-05,
+      "loss": 0.1575,
+      "step": 21829
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.3175591531755915e-05,
+      "loss": 0.1137,
+      "step": 21830
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.316313823163138e-05,
+      "loss": 0.1264,
+      "step": 21831
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.315068493150685e-05,
+      "loss": 0.1819,
+      "step": 21832
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.3138231631382306e-05,
+      "loss": 0.0923,
+      "step": 21833
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.312577833125778e-05,
+      "loss": 0.154,
+      "step": 21834
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.3113325031133245e-05,
+      "loss": 0.2412,
+      "step": 21835
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.310087173100871e-05,
+      "loss": 0.1294,
+      "step": 21836
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.308841843088418e-05,
+      "loss": 0.0953,
+      "step": 21837
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.307596513075965e-05,
+      "loss": 0.0881,
+      "step": 21838
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.3063511830635116e-05,
+      "loss": 0.1635,
+      "step": 21839
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.305105853051058e-05,
+      "loss": 0.1285,
+      "step": 21840
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.303860523038605e-05,
+      "loss": 0.1536,
+      "step": 21841
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.3026151930261513e-05,
+      "loss": 0.223,
+      "step": 21842
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.301369863013698e-05,
+      "loss": 0.1345,
+      "step": 21843
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.3001245330012445e-05,
+      "loss": 0.0971,
+      "step": 21844
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.298879202988792e-05,
+      "loss": 0.0866,
+      "step": 21845
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.2976338729763384e-05,
+      "loss": 0.0753,
+      "step": 21846
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.296388542963885e-05,
+      "loss": 0.1547,
+      "step": 21847
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.295143212951432e-05,
+      "loss": 0.1367,
+      "step": 21848
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.293897882938978e-05,
+      "loss": 0.1054,
+      "step": 21849
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 5.292652552926525e-05,
+      "loss": 0.1223,
+      "step": 21850
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.2914072229140714e-05,
+      "loss": 0.0925,
+      "step": 21851
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.2901618929016186e-05,
+      "loss": 0.1498,
+      "step": 21852
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.288916562889165e-05,
+      "loss": 0.1515,
+      "step": 21853
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.287671232876712e-05,
+      "loss": 0.2268,
+      "step": 21854
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.286425902864259e-05,
+      "loss": 0.1114,
+      "step": 21855
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.285180572851805e-05,
+      "loss": 0.0867,
+      "step": 21856
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.2839352428393516e-05,
+      "loss": 0.117,
+      "step": 21857
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.282689912826898e-05,
+      "loss": 0.1701,
+      "step": 21858
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.2814445828144455e-05,
+      "loss": 0.1473,
+      "step": 21859
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.280199252801992e-05,
+      "loss": 0.1704,
+      "step": 21860
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.278953922789539e-05,
+      "loss": 0.1176,
+      "step": 21861
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.277708592777086e-05,
+      "loss": 0.1935,
+      "step": 21862
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.2764632627646325e-05,
+      "loss": 0.1128,
+      "step": 21863
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.2752179327521785e-05,
+      "loss": 0.1317,
+      "step": 21864
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.273972602739725e-05,
+      "loss": 0.0914,
+      "step": 21865
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.272727272727272e-05,
+      "loss": 0.1842,
+      "step": 21866
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.271481942714819e-05,
+      "loss": 0.127,
+      "step": 21867
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.2702366127023655e-05,
+      "loss": 0.1307,
+      "step": 21868
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.268991282689913e-05,
+      "loss": 0.1774,
+      "step": 21869
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.2677459526774594e-05,
+      "loss": 0.1265,
+      "step": 21870
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.266500622665006e-05,
+      "loss": 0.0779,
+      "step": 21871
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.265255292652552e-05,
+      "loss": 0.1335,
+      "step": 21872
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.264009962640099e-05,
+      "loss": 0.2243,
+      "step": 21873
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.262764632627646e-05,
+      "loss": 0.1054,
+      "step": 21874
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.2615193026151924e-05,
+      "loss": 0.0966,
+      "step": 21875
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 5.2602739726027396e-05,
+      "loss": 0.0916,
+      "step": 21876
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.259028642590286e-05,
+      "loss": 0.0977,
+      "step": 21877
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.257783312577833e-05,
+      "loss": 0.1285,
+      "step": 21878
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.2565379825653794e-05,
+      "loss": 0.1197,
+      "step": 21879
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.255292652552926e-05,
+      "loss": 0.2349,
+      "step": 21880
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.2540473225404726e-05,
+      "loss": 0.0801,
+      "step": 21881
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.252801992528019e-05,
+      "loss": 0.079,
+      "step": 21882
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.251556662515566e-05,
+      "loss": 0.0832,
+      "step": 21883
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.250311332503113e-05,
+      "loss": 0.1106,
+      "step": 21884
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.2490660024906597e-05,
+      "loss": 0.1422,
+      "step": 21885
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.247820672478206e-05,
+      "loss": 0.1359,
+      "step": 21886
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.2465753424657535e-05,
+      "loss": 0.1052,
+      "step": 21887
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.2453300124532994e-05,
+      "loss": 0.0732,
+      "step": 21888
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.244084682440846e-05,
+      "loss": 0.0727,
+      "step": 21889
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.2428393524283926e-05,
+      "loss": 0.135,
+      "step": 21890
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.24159402241594e-05,
+      "loss": 0.2244,
+      "step": 21891
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.2403486924034865e-05,
+      "loss": 0.1407,
+      "step": 21892
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.239103362391033e-05,
+      "loss": 0.1255,
+      "step": 21893
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.2378580323785804e-05,
+      "loss": 0.1749,
+      "step": 21894
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.236612702366127e-05,
+      "loss": 0.1254,
+      "step": 21895
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.235367372353673e-05,
+      "loss": 0.0959,
+      "step": 21896
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.2341220423412195e-05,
+      "loss": 0.1125,
+      "step": 21897
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.232876712328767e-05,
+      "loss": 0.1971,
+      "step": 21898
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.231631382316313e-05,
+      "loss": 0.1757,
+      "step": 21899
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.23038605230386e-05,
+      "loss": 0.0977,
+      "step": 21900
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.229140722291407e-05,
+      "loss": 0.1968,
+      "step": 21901
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 5.227895392278954e-05,
+      "loss": 0.0972,
+      "step": 21902
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.2266500622665004e-05,
+      "loss": 0.0996,
+      "step": 21903
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.225404732254046e-05,
+      "loss": 0.1493,
+      "step": 21904
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.2241594022415936e-05,
+      "loss": 0.1307,
+      "step": 21905
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.22291407222914e-05,
+      "loss": 0.104,
+      "step": 21906
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.221668742216687e-05,
+      "loss": 0.111,
+      "step": 21907
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.220423412204234e-05,
+      "loss": 0.2106,
+      "step": 21908
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.2191780821917806e-05,
+      "loss": 0.0816,
+      "step": 21909
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.217932752179327e-05,
+      "loss": 0.3026,
+      "step": 21910
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.216687422166873e-05,
+      "loss": 0.1461,
+      "step": 21911
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.2154420921544204e-05,
+      "loss": 0.1869,
+      "step": 21912
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.214196762141967e-05,
+      "loss": 0.0933,
+      "step": 21913
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.2129514321295136e-05,
+      "loss": 0.0694,
+      "step": 21914
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.211706102117061e-05,
+      "loss": 0.165,
+      "step": 21915
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.2104607721046075e-05,
+      "loss": 0.1205,
+      "step": 21916
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.209215442092154e-05,
+      "loss": 0.2005,
+      "step": 21917
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.207970112079701e-05,
+      "loss": 0.1207,
+      "step": 21918
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.206724782067247e-05,
+      "loss": 0.1117,
+      "step": 21919
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.205479452054794e-05,
+      "loss": 0.1923,
+      "step": 21920
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.2042341220423404e-05,
+      "loss": 0.0962,
+      "step": 21921
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.202988792029888e-05,
+      "loss": 0.2908,
+      "step": 21922
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.201743462017434e-05,
+      "loss": 0.1961,
+      "step": 21923
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.200498132004981e-05,
+      "loss": 0.1747,
+      "step": 21924
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.1992528019925275e-05,
+      "loss": 0.1349,
+      "step": 21925
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.198007471980075e-05,
+      "loss": 0.0736,
+      "step": 21926
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.196762141967621e-05,
+      "loss": 0.1384,
+      "step": 21927
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 5.195516811955167e-05,
+      "loss": 0.1409,
+      "step": 21928
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.194271481942714e-05,
+      "loss": 0.141,
+      "step": 21929
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.193026151930261e-05,
+      "loss": 0.1303,
+      "step": 21930
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.191780821917808e-05,
+      "loss": 0.1926,
+      "step": 21931
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.1905354919053543e-05,
+      "loss": 0.0969,
+      "step": 21932
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.1892901618929016e-05,
+      "loss": 0.0973,
+      "step": 21933
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.188044831880448e-05,
+      "loss": 0.1674,
+      "step": 21934
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.186799501867994e-05,
+      "loss": 0.1897,
+      "step": 21935
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.185554171855541e-05,
+      "loss": 0.1894,
+      "step": 21936
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.184308841843088e-05,
+      "loss": 0.1142,
+      "step": 21937
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.1830635118306346e-05,
+      "loss": 0.083,
+      "step": 21938
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.181818181818181e-05,
+      "loss": 0.0985,
+      "step": 21939
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.1805728518057285e-05,
+      "loss": 0.0989,
+      "step": 21940
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.179327521793275e-05,
+      "loss": 0.1478,
+      "step": 21941
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.1780821917808216e-05,
+      "loss": 0.1102,
+      "step": 21942
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.1768368617683676e-05,
+      "loss": 0.1472,
+      "step": 21943
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.175591531755915e-05,
+      "loss": 0.1383,
+      "step": 21944
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.1743462017434614e-05,
+      "loss": 0.0836,
+      "step": 21945
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.173100871731008e-05,
+      "loss": 0.1454,
+      "step": 21946
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.171855541718555e-05,
+      "loss": 0.0951,
+      "step": 21947
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.170610211706102e-05,
+      "loss": 0.125,
+      "step": 21948
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.1693648816936485e-05,
+      "loss": 0.0976,
+      "step": 21949
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.168119551681196e-05,
+      "loss": 0.1702,
+      "step": 21950
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.166874221668742e-05,
+      "loss": 0.1511,
+      "step": 21951
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.165628891656288e-05,
+      "loss": 0.0695,
+      "step": 21952
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.164383561643835e-05,
+      "loss": 0.1701,
+      "step": 21953
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 5.163138231631382e-05,
+      "loss": 0.3276,
+      "step": 21954
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.161892901618929e-05,
+      "loss": 0.1329,
+      "step": 21955
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.160647571606475e-05,
+      "loss": 0.1117,
+      "step": 21956
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.1594022415940226e-05,
+      "loss": 0.0991,
+      "step": 21957
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.158156911581569e-05,
+      "loss": 0.0602,
+      "step": 21958
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.156911581569115e-05,
+      "loss": 0.1454,
+      "step": 21959
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.155666251556662e-05,
+      "loss": 0.1623,
+      "step": 21960
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.154420921544209e-05,
+      "loss": 0.1716,
+      "step": 21961
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.1531755915317556e-05,
+      "loss": 0.1668,
+      "step": 21962
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.151930261519302e-05,
+      "loss": 0.0629,
+      "step": 21963
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.150684931506849e-05,
+      "loss": 0.1189,
+      "step": 21964
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.149439601494396e-05,
+      "loss": 0.1273,
+      "step": 21965
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.148194271481942e-05,
+      "loss": 0.1773,
+      "step": 21966
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.1469489414694885e-05,
+      "loss": 0.1474,
+      "step": 21967
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.145703611457035e-05,
+      "loss": 0.2126,
+      "step": 21968
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.1444582814445824e-05,
+      "loss": 0.1726,
+      "step": 21969
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.143212951432129e-05,
+      "loss": 0.1614,
+      "step": 21970
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.1419676214196756e-05,
+      "loss": 0.0846,
+      "step": 21971
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.140722291407223e-05,
+      "loss": 0.2041,
+      "step": 21972
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.1394769613947695e-05,
+      "loss": 0.1202,
+      "step": 21973
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.1382316313823154e-05,
+      "loss": 0.141,
+      "step": 21974
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.136986301369862e-05,
+      "loss": 0.1132,
+      "step": 21975
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.135740971357409e-05,
+      "loss": 0.0895,
+      "step": 21976
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.134495641344956e-05,
+      "loss": 0.1292,
+      "step": 21977
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.1332503113325024e-05,
+      "loss": 0.0987,
+      "step": 21978
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.13200498132005e-05,
+      "loss": 0.133,
+      "step": 21979
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 5.130759651307596e-05,
+      "loss": 0.1639,
+      "step": 21980
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.129514321295143e-05,
+      "loss": 0.1112,
+      "step": 21981
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.128268991282689e-05,
+      "loss": 0.1526,
+      "step": 21982
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.127023661270236e-05,
+      "loss": 0.0657,
+      "step": 21983
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.125778331257783e-05,
+      "loss": 0.1838,
+      "step": 21984
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.124533001245329e-05,
+      "loss": 0.2029,
+      "step": 21985
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.1232876712328765e-05,
+      "loss": 0.1283,
+      "step": 21986
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.122042341220423e-05,
+      "loss": 0.0956,
+      "step": 21987
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.12079701120797e-05,
+      "loss": 0.2485,
+      "step": 21988
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.119551681195517e-05,
+      "loss": 0.1205,
+      "step": 21989
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.118306351183063e-05,
+      "loss": 0.3274,
+      "step": 21990
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.1170610211706095e-05,
+      "loss": 0.1686,
+      "step": 21991
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.115815691158156e-05,
+      "loss": 0.1261,
+      "step": 21992
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.1145703611457034e-05,
+      "loss": 0.0983,
+      "step": 21993
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.11332503113325e-05,
+      "loss": 0.115,
+      "step": 21994
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.1120797011207966e-05,
+      "loss": 0.1349,
+      "step": 21995
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.110834371108344e-05,
+      "loss": 0.0941,
+      "step": 21996
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.1095890410958904e-05,
+      "loss": 0.198,
+      "step": 21997
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.1083437110834364e-05,
+      "loss": 0.1532,
+      "step": 21998
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.107098381070983e-05,
+      "loss": 0.1105,
+      "step": 21999
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.10585305105853e-05,
+      "loss": 0.0727,
+      "step": 22000
+    },
+    {
+      "epoch": 8.43,
+      "eval_cer": 0.10087207600913047,
+      "eval_loss": 0.39484187960624695,
+      "eval_runtime": 47.5765,
+      "eval_samples_per_second": 36.846,
+      "eval_steps_per_second": 4.624,
+      "eval_wer": 0.1607231750916414,
+      "step": 22000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.104607721046077e-05,
+      "loss": 0.1333,
+      "step": 22001
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.1033623910336234e-05,
+      "loss": 0.0704,
+      "step": 22002
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.102117061021171e-05,
+      "loss": 0.1968,
+      "step": 22003
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.100871731008717e-05,
+      "loss": 0.202,
+      "step": 22004
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.099626400996264e-05,
+      "loss": 0.2039,
+      "step": 22005
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 5.09838107098381e-05,
+      "loss": 0.289,
+      "step": 22006
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.097135740971357e-05,
+      "loss": 0.1601,
+      "step": 22007
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.0958904109589037e-05,
+      "loss": 0.0729,
+      "step": 22008
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.09464508094645e-05,
+      "loss": 0.2465,
+      "step": 22009
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.093399750933997e-05,
+      "loss": 0.2098,
+      "step": 22010
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.092154420921544e-05,
+      "loss": 0.1114,
+      "step": 22011
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.090909090909091e-05,
+      "loss": 0.1705,
+      "step": 22012
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.089663760896637e-05,
+      "loss": 0.0964,
+      "step": 22013
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.088418430884183e-05,
+      "loss": 0.0983,
+      "step": 22014
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.0871731008717305e-05,
+      "loss": 0.1734,
+      "step": 22015
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.085927770859277e-05,
+      "loss": 0.1776,
+      "step": 22016
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.084682440846824e-05,
+      "loss": 0.1137,
+      "step": 22017
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.083437110834371e-05,
+      "loss": 0.1077,
+      "step": 22018
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.0821917808219175e-05,
+      "loss": 0.1036,
+      "step": 22019
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.080946450809464e-05,
+      "loss": 0.136,
+      "step": 22020
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.07970112079701e-05,
+      "loss": 0.1368,
+      "step": 22021
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.078455790784557e-05,
+      "loss": 0.2518,
+      "step": 22022
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.077210460772104e-05,
+      "loss": 0.1262,
+      "step": 22023
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.0759651307596505e-05,
+      "loss": 0.1369,
+      "step": 22024
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.074719800747198e-05,
+      "loss": 0.1333,
+      "step": 22025
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.0734744707347444e-05,
+      "loss": 0.1258,
+      "step": 22026
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.072229140722291e-05,
+      "loss": 0.1568,
+      "step": 22027
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.070983810709838e-05,
+      "loss": 0.158,
+      "step": 22028
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.069738480697384e-05,
+      "loss": 0.2199,
+      "step": 22029
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.068493150684931e-05,
+      "loss": 0.1188,
+      "step": 22030
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.0672478206724774e-05,
+      "loss": 0.1182,
+      "step": 22031
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.0660024906600246e-05,
+      "loss": 0.102,
+      "step": 22032
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 5.064757160647571e-05,
+      "loss": 0.1934,
+      "step": 22033
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.063511830635118e-05,
+      "loss": 0.0803,
+      "step": 22034
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.062266500622665e-05,
+      "loss": 0.2225,
+      "step": 22035
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.061021170610212e-05,
+      "loss": 0.1966,
+      "step": 22036
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.0597758405977576e-05,
+      "loss": 0.2447,
+      "step": 22037
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.058530510585304e-05,
+      "loss": 0.1541,
+      "step": 22038
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.0572851805728515e-05,
+      "loss": 0.1811,
+      "step": 22039
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.056039850560398e-05,
+      "loss": 0.1382,
+      "step": 22040
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.0547945205479447e-05,
+      "loss": 0.1763,
+      "step": 22041
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.053549190535492e-05,
+      "loss": 0.1027,
+      "step": 22042
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.0523038605230385e-05,
+      "loss": 0.1292,
+      "step": 22043
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.051058530510585e-05,
+      "loss": 0.1323,
+      "step": 22044
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.049813200498131e-05,
+      "loss": 0.1152,
+      "step": 22045
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.048567870485678e-05,
+      "loss": 0.0947,
+      "step": 22046
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.047322540473225e-05,
+      "loss": 0.134,
+      "step": 22047
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.0460772104607715e-05,
+      "loss": 0.1841,
+      "step": 22048
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.044831880448318e-05,
+      "loss": 0.1428,
+      "step": 22049
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.0435865504358654e-05,
+      "loss": 0.0993,
+      "step": 22050
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.042341220423412e-05,
+      "loss": 0.0785,
+      "step": 22051
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.0410958904109586e-05,
+      "loss": 0.1632,
+      "step": 22052
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.039850560398505e-05,
+      "loss": 0.1842,
+      "step": 22053
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.038605230386052e-05,
+      "loss": 0.28,
+      "step": 22054
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.0373599003735983e-05,
+      "loss": 0.1899,
+      "step": 22055
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.036114570361145e-05,
+      "loss": 0.1642,
+      "step": 22056
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.034869240348692e-05,
+      "loss": 0.0938,
+      "step": 22057
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.033623910336239e-05,
+      "loss": 0.0686,
+      "step": 22058
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 5.0323785803237854e-05,
+      "loss": 0.1225,
+      "step": 22059
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.031133250311333e-05,
+      "loss": 0.2223,
+      "step": 22060
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.0298879202988786e-05,
+      "loss": 0.1121,
+      "step": 22061
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.028642590286425e-05,
+      "loss": 0.1719,
+      "step": 22062
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.027397260273972e-05,
+      "loss": 0.0968,
+      "step": 22063
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.026151930261519e-05,
+      "loss": 0.0866,
+      "step": 22064
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.0249066002490656e-05,
+      "loss": 0.167,
+      "step": 22065
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.023661270236612e-05,
+      "loss": 0.2561,
+      "step": 22066
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.0224159402241595e-05,
+      "loss": 0.1579,
+      "step": 22067
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.0211706102117054e-05,
+      "loss": 0.2173,
+      "step": 22068
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.019925280199252e-05,
+      "loss": 0.107,
+      "step": 22069
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.0186799501867986e-05,
+      "loss": 0.119,
+      "step": 22070
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.017434620174346e-05,
+      "loss": 0.0763,
+      "step": 22071
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.0161892901618925e-05,
+      "loss": 0.2608,
+      "step": 22072
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.014943960149439e-05,
+      "loss": 0.2016,
+      "step": 22073
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.0136986301369863e-05,
+      "loss": 0.1388,
+      "step": 22074
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.012453300124533e-05,
+      "loss": 0.1595,
+      "step": 22075
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.011207970112079e-05,
+      "loss": 0.0901,
+      "step": 22076
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.0099626400996255e-05,
+      "loss": 0.1165,
+      "step": 22077
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.008717310087173e-05,
+      "loss": 0.2097,
+      "step": 22078
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.007471980074719e-05,
+      "loss": 0.1952,
+      "step": 22079
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.006226650062266e-05,
+      "loss": 0.1277,
+      "step": 22080
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.004981320049813e-05,
+      "loss": 0.1367,
+      "step": 22081
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.00373599003736e-05,
+      "loss": 0.1662,
+      "step": 22082
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.0024906600249064e-05,
+      "loss": 0.1158,
+      "step": 22083
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 5.001245330012452e-05,
+      "loss": 0.1692,
+      "step": 22084
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 4.9999999999999996e-05,
+      "loss": 0.1654,
+      "step": 22085
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.998754669987546e-05,
+      "loss": 0.1697,
+      "step": 22086
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.997509339975093e-05,
+      "loss": 0.1373,
+      "step": 22087
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.99626400996264e-05,
+      "loss": 0.1279,
+      "step": 22088
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.9950186799501866e-05,
+      "loss": 0.121,
+      "step": 22089
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.993773349937733e-05,
+      "loss": 0.0589,
+      "step": 22090
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.99252801992528e-05,
+      "loss": 0.1311,
+      "step": 22091
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.9912826899128264e-05,
+      "loss": 0.1182,
+      "step": 22092
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.990037359900373e-05,
+      "loss": 0.1821,
+      "step": 22093
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.9887920298879196e-05,
+      "loss": 0.1086,
+      "step": 22094
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.987546699875466e-05,
+      "loss": 0.1125,
+      "step": 22095
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.9863013698630135e-05,
+      "loss": 0.1462,
+      "step": 22096
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.98505603985056e-05,
+      "loss": 0.1937,
+      "step": 22097
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.9838107098381066e-05,
+      "loss": 0.2626,
+      "step": 22098
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.982565379825654e-05,
+      "loss": 0.1674,
+      "step": 22099
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.9813200498132e-05,
+      "loss": 0.1007,
+      "step": 22100
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.9800747198007464e-05,
+      "loss": 0.1159,
+      "step": 22101
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.978829389788293e-05,
+      "loss": 0.0968,
+      "step": 22102
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.97758405977584e-05,
+      "loss": 0.2484,
+      "step": 22103
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.976338729763387e-05,
+      "loss": 0.1985,
+      "step": 22104
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.9750933997509335e-05,
+      "loss": 0.2129,
+      "step": 22105
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.973848069738481e-05,
+      "loss": 0.0777,
+      "step": 22106
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.9726027397260274e-05,
+      "loss": 0.0855,
+      "step": 22107
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.971357409713573e-05,
+      "loss": 0.1445,
+      "step": 22108
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.97011207970112e-05,
+      "loss": 0.1125,
+      "step": 22109
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.968866749688667e-05,
+      "loss": 0.1749,
+      "step": 22110
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 4.967621419676214e-05,
+      "loss": 0.1865,
+      "step": 22111
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.96637608966376e-05,
+      "loss": 0.1165,
+      "step": 22112
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.9651307596513076e-05,
+      "loss": 0.097,
+      "step": 22113
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.963885429638854e-05,
+      "loss": 0.1078,
+      "step": 22114
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.962640099626401e-05,
+      "loss": 0.0796,
+      "step": 22115
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.961394769613947e-05,
+      "loss": 0.1746,
+      "step": 22116
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.960149439601494e-05,
+      "loss": 0.1431,
+      "step": 22117
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.9589041095890406e-05,
+      "loss": 0.2204,
+      "step": 22118
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.957658779576587e-05,
+      "loss": 0.1345,
+      "step": 22119
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.9564134495641344e-05,
+      "loss": 0.0912,
+      "step": 22120
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.955168119551681e-05,
+      "loss": 0.0912,
+      "step": 22121
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.9539227895392276e-05,
+      "loss": 0.1502,
+      "step": 22122
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.9526774595267735e-05,
+      "loss": 0.1642,
+      "step": 22123
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.951432129514321e-05,
+      "loss": 0.1315,
+      "step": 22124
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.9501867995018674e-05,
+      "loss": 0.1022,
+      "step": 22125
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.948941469489414e-05,
+      "loss": 0.1197,
+      "step": 22126
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.947696139476961e-05,
+      "loss": 0.1661,
+      "step": 22127
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.946450809464508e-05,
+      "loss": 0.1781,
+      "step": 22128
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.9452054794520545e-05,
+      "loss": 0.2017,
+      "step": 22129
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.943960149439601e-05,
+      "loss": 0.1155,
+      "step": 22130
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.9427148194271477e-05,
+      "loss": 0.1281,
+      "step": 22131
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.941469489414694e-05,
+      "loss": 0.1655,
+      "step": 22132
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.940224159402241e-05,
+      "loss": 0.1743,
+      "step": 22133
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.938978829389788e-05,
+      "loss": 0.2032,
+      "step": 22134
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.937733499377335e-05,
+      "loss": 0.2439,
+      "step": 22135
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.936488169364881e-05,
+      "loss": 0.1161,
+      "step": 22136
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 4.935242839352428e-05,
+      "loss": 0.1674,
+      "step": 22137
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.933997509339975e-05,
+      "loss": 0.1289,
+      "step": 22138
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.932752179327521e-05,
+      "loss": 0.1376,
+      "step": 22139
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.931506849315068e-05,
+      "loss": 0.1213,
+      "step": 22140
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.930261519302614e-05,
+      "loss": 0.1919,
+      "step": 22141
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.9290161892901615e-05,
+      "loss": 0.1423,
+      "step": 22142
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.927770859277708e-05,
+      "loss": 0.2031,
+      "step": 22143
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.926525529265255e-05,
+      "loss": 0.1113,
+      "step": 22144
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.925280199252802e-05,
+      "loss": 0.0824,
+      "step": 22145
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.9240348692403486e-05,
+      "loss": 0.1129,
+      "step": 22146
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.9227895392278945e-05,
+      "loss": 0.1278,
+      "step": 22147
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.921544209215441e-05,
+      "loss": 0.1331,
+      "step": 22148
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.9202988792029884e-05,
+      "loss": 0.1502,
+      "step": 22149
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.919053549190535e-05,
+      "loss": 0.1576,
+      "step": 22150
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.9178082191780816e-05,
+      "loss": 0.1035,
+      "step": 22151
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.916562889165629e-05,
+      "loss": 0.104,
+      "step": 22152
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.9153175591531754e-05,
+      "loss": 0.1977,
+      "step": 22153
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.914072229140722e-05,
+      "loss": 0.2124,
+      "step": 22154
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.912826899128268e-05,
+      "loss": 0.1613,
+      "step": 22155
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.911581569115815e-05,
+      "loss": 0.124,
+      "step": 22156
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.910336239103362e-05,
+      "loss": 0.1204,
+      "step": 22157
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.9090909090909084e-05,
+      "loss": 0.0898,
+      "step": 22158
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.907845579078456e-05,
+      "loss": 0.1187,
+      "step": 22159
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.906600249066002e-05,
+      "loss": 0.1612,
+      "step": 22160
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.905354919053549e-05,
+      "loss": 0.2058,
+      "step": 22161
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.904109589041096e-05,
+      "loss": 0.1342,
+      "step": 22162
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 4.902864259028642e-05,
+      "loss": 0.1036,
+      "step": 22163
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.9016189290161887e-05,
+      "loss": 0.1082,
+      "step": 22164
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.900373599003735e-05,
+      "loss": 0.2511,
+      "step": 22165
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.8991282689912825e-05,
+      "loss": 0.2019,
+      "step": 22166
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.897882938978829e-05,
+      "loss": 0.1931,
+      "step": 22167
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.896637608966376e-05,
+      "loss": 0.1426,
+      "step": 22168
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.895392278953923e-05,
+      "loss": 0.101,
+      "step": 22169
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.8941469489414696e-05,
+      "loss": 0.0875,
+      "step": 22170
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.8929016189290155e-05,
+      "loss": 0.2392,
+      "step": 22171
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.891656288916562e-05,
+      "loss": 0.2482,
+      "step": 22172
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.8904109589041094e-05,
+      "loss": 0.1514,
+      "step": 22173
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.889165628891656e-05,
+      "loss": 0.2155,
+      "step": 22174
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.8879202988792026e-05,
+      "loss": 0.1111,
+      "step": 22175
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.886674968866749e-05,
+      "loss": 0.1743,
+      "step": 22176
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.8854296388542964e-05,
+      "loss": 0.1034,
+      "step": 22177
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.884184308841842e-05,
+      "loss": 0.0622,
+      "step": 22178
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.882938978829389e-05,
+      "loss": 0.1572,
+      "step": 22179
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.881693648816936e-05,
+      "loss": 0.1345,
+      "step": 22180
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.880448318804483e-05,
+      "loss": 0.2086,
+      "step": 22181
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.8792029887920294e-05,
+      "loss": 0.1082,
+      "step": 22182
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.877957658779576e-05,
+      "loss": 0.1573,
+      "step": 22183
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.876712328767123e-05,
+      "loss": 0.0963,
+      "step": 22184
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.87546699875467e-05,
+      "loss": 0.255,
+      "step": 22185
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.874221668742216e-05,
+      "loss": 0.1194,
+      "step": 22186
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.8729763387297624e-05,
+      "loss": 0.1297,
+      "step": 22187
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.8717310087173096e-05,
+      "loss": 0.1414,
+      "step": 22188
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 4.870485678704856e-05,
+      "loss": 0.0653,
+      "step": 22189
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.869240348692403e-05,
+      "loss": 0.1188,
+      "step": 22190
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.86799501867995e-05,
+      "loss": 0.1822,
+      "step": 22191
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.866749688667497e-05,
+      "loss": 0.141,
+      "step": 22192
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.865504358655043e-05,
+      "loss": 0.1897,
+      "step": 22193
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.864259028642589e-05,
+      "loss": 0.1446,
+      "step": 22194
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.8630136986301365e-05,
+      "loss": 0.1787,
+      "step": 22195
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.861768368617683e-05,
+      "loss": 0.1221,
+      "step": 22196
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.86052303860523e-05,
+      "loss": 0.1565,
+      "step": 22197
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.859277708592777e-05,
+      "loss": 0.1928,
+      "step": 22198
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.8580323785803235e-05,
+      "loss": 0.1096,
+      "step": 22199
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.85678704856787e-05,
+      "loss": 0.0833,
+      "step": 22200
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.8555417185554174e-05,
+      "loss": 0.1624,
+      "step": 22201
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.854296388542963e-05,
+      "loss": 0.0695,
+      "step": 22202
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.85305105853051e-05,
+      "loss": 0.2047,
+      "step": 22203
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.8518057285180565e-05,
+      "loss": 0.1957,
+      "step": 22204
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.850560398505604e-05,
+      "loss": 0.1132,
+      "step": 22205
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.8493150684931504e-05,
+      "loss": 0.1197,
+      "step": 22206
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.848069738480697e-05,
+      "loss": 0.1298,
+      "step": 22207
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.846824408468244e-05,
+      "loss": 0.0801,
+      "step": 22208
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.845579078455791e-05,
+      "loss": 0.098,
+      "step": 22209
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.844333748443337e-05,
+      "loss": 0.2182,
+      "step": 22210
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.8430884184308833e-05,
+      "loss": 0.1474,
+      "step": 22211
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.8418430884184306e-05,
+      "loss": 0.1507,
+      "step": 22212
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.840597758405977e-05,
+      "loss": 0.1106,
+      "step": 22213
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.839352428393524e-05,
+      "loss": 0.1524,
+      "step": 22214
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 4.838107098381071e-05,
+      "loss": 0.0842,
+      "step": 22215
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.836861768368618e-05,
+      "loss": 0.1243,
+      "step": 22216
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.835616438356164e-05,
+      "loss": 0.0978,
+      "step": 22217
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.83437110834371e-05,
+      "loss": 0.1794,
+      "step": 22218
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.8331257783312575e-05,
+      "loss": 0.1505,
+      "step": 22219
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.831880448318804e-05,
+      "loss": 0.0994,
+      "step": 22220
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.8306351183063506e-05,
+      "loss": 0.1421,
+      "step": 22221
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.829389788293897e-05,
+      "loss": 0.149,
+      "step": 22222
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.8281444582814445e-05,
+      "loss": 0.1909,
+      "step": 22223
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.826899128268991e-05,
+      "loss": 0.1518,
+      "step": 22224
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.825653798256538e-05,
+      "loss": 0.1487,
+      "step": 22225
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.8244084682440836e-05,
+      "loss": 0.1926,
+      "step": 22226
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.823163138231631e-05,
+      "loss": 0.1043,
+      "step": 22227
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.8219178082191775e-05,
+      "loss": 0.1081,
+      "step": 22228
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.820672478206724e-05,
+      "loss": 0.2895,
+      "step": 22229
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.8194271481942714e-05,
+      "loss": 0.1572,
+      "step": 22230
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.818181818181818e-05,
+      "loss": 0.2338,
+      "step": 22231
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.8169364881693645e-05,
+      "loss": 0.1567,
+      "step": 22232
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.8156911581569105e-05,
+      "loss": 0.0914,
+      "step": 22233
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.814445828144458e-05,
+      "loss": 0.0749,
+      "step": 22234
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.813200498132004e-05,
+      "loss": 0.2435,
+      "step": 22235
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.811955168119551e-05,
+      "loss": 0.1444,
+      "step": 22236
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.810709838107098e-05,
+      "loss": 0.1512,
+      "step": 22237
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.809464508094645e-05,
+      "loss": 0.1116,
+      "step": 22238
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.8082191780821914e-05,
+      "loss": 0.1416,
+      "step": 22239
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.8069738480697387e-05,
+      "loss": 0.0866,
+      "step": 22240
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 4.8057285180572846e-05,
+      "loss": 0.2731,
+      "step": 22241
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.804483188044831e-05,
+      "loss": 0.1223,
+      "step": 22242
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.803237858032378e-05,
+      "loss": 0.2371,
+      "step": 22243
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.801992528019925e-05,
+      "loss": 0.0933,
+      "step": 22244
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.8007471980074716e-05,
+      "loss": 0.1005,
+      "step": 22245
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.799501867995018e-05,
+      "loss": 0.0785,
+      "step": 22246
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.7982565379825655e-05,
+      "loss": 0.1721,
+      "step": 22247
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.797011207970112e-05,
+      "loss": 0.1638,
+      "step": 22248
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.795765877957658e-05,
+      "loss": 0.1733,
+      "step": 22249
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.7945205479452046e-05,
+      "loss": 0.175,
+      "step": 22250
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.793275217932752e-05,
+      "loss": 0.1638,
+      "step": 22251
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.7920298879202985e-05,
+      "loss": 0.09,
+      "step": 22252
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.790784557907845e-05,
+      "loss": 0.1336,
+      "step": 22253
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.789539227895392e-05,
+      "loss": 0.2611,
+      "step": 22254
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.788293897882939e-05,
+      "loss": 0.1231,
+      "step": 22255
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.7870485678704855e-05,
+      "loss": 0.0961,
+      "step": 22256
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.7858032378580314e-05,
+      "loss": 0.0825,
+      "step": 22257
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.784557907845579e-05,
+      "loss": 0.1023,
+      "step": 22258
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.783312577833125e-05,
+      "loss": 0.0889,
+      "step": 22259
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.782067247820672e-05,
+      "loss": 0.2254,
+      "step": 22260
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.7808219178082185e-05,
+      "loss": 0.1501,
+      "step": 22261
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.779576587795766e-05,
+      "loss": 0.1342,
+      "step": 22262
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.7783312577833124e-05,
+      "loss": 0.116,
+      "step": 22263
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.777085927770859e-05,
+      "loss": 0.0885,
+      "step": 22264
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.7758405977584055e-05,
+      "loss": 0.1805,
+      "step": 22265
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.774595267745952e-05,
+      "loss": 0.2171,
+      "step": 22266
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 4.773349937733499e-05,
+      "loss": 0.166,
+      "step": 22267
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.772104607721045e-05,
+      "loss": 0.1165,
+      "step": 22268
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.7708592777085926e-05,
+      "loss": 0.0703,
+      "step": 22269
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.769613947696139e-05,
+      "loss": 0.193,
+      "step": 22270
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.768368617683686e-05,
+      "loss": 0.0866,
+      "step": 22271
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.767123287671233e-05,
+      "loss": 0.2181,
+      "step": 22272
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.765877957658779e-05,
+      "loss": 0.1301,
+      "step": 22273
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.7646326276463256e-05,
+      "loss": 0.1234,
+      "step": 22274
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.763387297633872e-05,
+      "loss": 0.0932,
+      "step": 22275
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.7621419676214194e-05,
+      "loss": 0.2295,
+      "step": 22276
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.760896637608966e-05,
+      "loss": 0.1178,
+      "step": 22277
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.7596513075965126e-05,
+      "loss": 0.2074,
+      "step": 22278
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.75840597758406e-05,
+      "loss": 0.1832,
+      "step": 22279
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.757160647571606e-05,
+      "loss": 0.1976,
+      "step": 22280
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.7559153175591524e-05,
+      "loss": 0.1313,
+      "step": 22281
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.754669987546699e-05,
+      "loss": 0.2069,
+      "step": 22282
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.753424657534246e-05,
+      "loss": 0.1093,
+      "step": 22283
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.752179327521793e-05,
+      "loss": 0.2622,
+      "step": 22284
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.7509339975093395e-05,
+      "loss": 0.2251,
+      "step": 22285
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.749688667496887e-05,
+      "loss": 0.1454,
+      "step": 22286
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.748443337484433e-05,
+      "loss": 0.1388,
+      "step": 22287
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.747198007471979e-05,
+      "loss": 0.079,
+      "step": 22288
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.745952677459526e-05,
+      "loss": 0.0924,
+      "step": 22289
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.744707347447073e-05,
+      "loss": 0.1898,
+      "step": 22290
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.74346201743462e-05,
+      "loss": 0.1373,
+      "step": 22291
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.742216687422166e-05,
+      "loss": 0.1819,
+      "step": 22292
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 4.7409713574097136e-05,
+      "loss": 0.1171,
+      "step": 22293
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.73972602739726e-05,
+      "loss": 0.1301,
+      "step": 22294
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.738480697384807e-05,
+      "loss": 0.0752,
+      "step": 22295
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.737235367372353e-05,
+      "loss": 0.0879,
+      "step": 22296
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.7359900373599e-05,
+      "loss": 0.2796,
+      "step": 22297
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.7347447073474466e-05,
+      "loss": 0.2303,
+      "step": 22298
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.733499377334993e-05,
+      "loss": 0.1664,
+      "step": 22299
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.7322540473225404e-05,
+      "loss": 0.1028,
+      "step": 22300
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.731008717310087e-05,
+      "loss": 0.0847,
+      "step": 22301
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.7297633872976336e-05,
+      "loss": 0.1185,
+      "step": 22302
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.72851805728518e-05,
+      "loss": 0.0989,
+      "step": 22303
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.727272727272727e-05,
+      "loss": 0.2488,
+      "step": 22304
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.7260273972602734e-05,
+      "loss": 0.2062,
+      "step": 22305
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.72478206724782e-05,
+      "loss": 0.1421,
+      "step": 22306
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.7235367372353666e-05,
+      "loss": 0.1427,
+      "step": 22307
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.722291407222914e-05,
+      "loss": 0.0868,
+      "step": 22308
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.7210460772104604e-05,
+      "loss": 0.136,
+      "step": 22309
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.719800747198007e-05,
+      "loss": 0.2709,
+      "step": 22310
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.718555417185554e-05,
+      "loss": 0.1773,
+      "step": 22311
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.7173100871731e-05,
+      "loss": 0.153,
+      "step": 22312
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.716064757160647e-05,
+      "loss": 0.096,
+      "step": 22313
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.7148194271481934e-05,
+      "loss": 0.1607,
+      "step": 22314
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.713574097135741e-05,
+      "loss": 0.0923,
+      "step": 22315
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.712328767123287e-05,
+      "loss": 0.2995,
+      "step": 22316
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.711083437110834e-05,
+      "loss": 0.1269,
+      "step": 22317
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.709838107098381e-05,
+      "loss": 0.1153,
+      "step": 22318
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 4.708592777085928e-05,
+      "loss": 0.1282,
+      "step": 22319
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.707347447073474e-05,
+      "loss": 0.1049,
+      "step": 22320
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.70610211706102e-05,
+      "loss": 0.0861,
+      "step": 22321
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.7048567870485675e-05,
+      "loss": 0.1503,
+      "step": 22322
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.703611457036114e-05,
+      "loss": 0.1344,
+      "step": 22323
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.702366127023661e-05,
+      "loss": 0.141,
+      "step": 22324
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.701120797011208e-05,
+      "loss": 0.0954,
+      "step": 22325
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.6998754669987546e-05,
+      "loss": 0.1101,
+      "step": 22326
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.698630136986301e-05,
+      "loss": 0.1416,
+      "step": 22327
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.697384806973847e-05,
+      "loss": 0.086,
+      "step": 22328
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.6961394769613944e-05,
+      "loss": 0.3042,
+      "step": 22329
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.694894146948941e-05,
+      "loss": 0.228,
+      "step": 22330
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.6936488169364876e-05,
+      "loss": 0.1251,
+      "step": 22331
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.692403486924035e-05,
+      "loss": 0.1054,
+      "step": 22332
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.6911581569115814e-05,
+      "loss": 0.0831,
+      "step": 22333
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.689912826899128e-05,
+      "loss": 0.2004,
+      "step": 22334
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.688667496886674e-05,
+      "loss": 0.2676,
+      "step": 22335
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.687422166874221e-05,
+      "loss": 0.1797,
+      "step": 22336
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.686176836861768e-05,
+      "loss": 0.1542,
+      "step": 22337
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.6849315068493144e-05,
+      "loss": 0.1541,
+      "step": 22338
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.683686176836862e-05,
+      "loss": 0.1017,
+      "step": 22339
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.682440846824408e-05,
+      "loss": 0.0749,
+      "step": 22340
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.681195516811955e-05,
+      "loss": 0.1091,
+      "step": 22341
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.6799501867995015e-05,
+      "loss": 0.1956,
+      "step": 22342
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.678704856787048e-05,
+      "loss": 0.1147,
+      "step": 22343
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.6774595267745946e-05,
+      "loss": 0.102,
+      "step": 22344
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.676214196762141e-05,
+      "loss": 0.1059,
+      "step": 22345
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 4.6749688667496885e-05,
+      "loss": 0.1289,
+      "step": 22346
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.673723536737235e-05,
+      "loss": 0.1999,
+      "step": 22347
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.672478206724782e-05,
+      "loss": 0.218,
+      "step": 22348
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.671232876712328e-05,
+      "loss": 0.1505,
+      "step": 22349
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.6699875466998756e-05,
+      "loss": 0.1543,
+      "step": 22350
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.6687422166874215e-05,
+      "loss": 0.1037,
+      "step": 22351
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.667496886674968e-05,
+      "loss": 0.1019,
+      "step": 22352
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.666251556662515e-05,
+      "loss": 0.1838,
+      "step": 22353
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.665006226650062e-05,
+      "loss": 0.1946,
+      "step": 22354
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.6637608966376085e-05,
+      "loss": 0.2016,
+      "step": 22355
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.662515566625155e-05,
+      "loss": 0.1107,
+      "step": 22356
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.6612702366127024e-05,
+      "loss": 0.1082,
+      "step": 22357
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.660024906600249e-05,
+      "loss": 0.1077,
+      "step": 22358
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.658779576587795e-05,
+      "loss": 0.2237,
+      "step": 22359
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.6575342465753415e-05,
+      "loss": 0.1792,
+      "step": 22360
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.656288916562889e-05,
+      "loss": 0.1747,
+      "step": 22361
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.6550435865504354e-05,
+      "loss": 0.1125,
+      "step": 22362
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.653798256537982e-05,
+      "loss": 0.0829,
+      "step": 22363
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.652552926525529e-05,
+      "loss": 0.1298,
+      "step": 22364
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.651307596513076e-05,
+      "loss": 0.1577,
+      "step": 22365
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.6500622665006224e-05,
+      "loss": 0.213,
+      "step": 22366
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.6488169364881683e-05,
+      "loss": 0.1979,
+      "step": 22367
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.6475716064757156e-05,
+      "loss": 0.1166,
+      "step": 22368
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.646326276463262e-05,
+      "loss": 0.0995,
+      "step": 22369
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.645080946450809e-05,
+      "loss": 0.079,
+      "step": 22370
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.643835616438356e-05,
+      "loss": 0.1487,
+      "step": 22371
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 4.642590286425903e-05,
+      "loss": 0.2004,
+      "step": 22372
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.641344956413449e-05,
+      "loss": 0.2389,
+      "step": 22373
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6400996264009965e-05,
+      "loss": 0.1111,
+      "step": 22374
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6388542963885425e-05,
+      "loss": 0.1379,
+      "step": 22375
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.637608966376089e-05,
+      "loss": 0.0974,
+      "step": 22376
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6363636363636356e-05,
+      "loss": 0.1351,
+      "step": 22377
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.635118306351183e-05,
+      "loss": 0.131,
+      "step": 22378
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6338729763387295e-05,
+      "loss": 0.1577,
+      "step": 22379
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.632627646326276e-05,
+      "loss": 0.2651,
+      "step": 22380
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6313823163138234e-05,
+      "loss": 0.1447,
+      "step": 22381
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.63013698630137e-05,
+      "loss": 0.1183,
+      "step": 22382
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.628891656288916e-05,
+      "loss": 0.0851,
+      "step": 22383
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6276463262764625e-05,
+      "loss": 0.1201,
+      "step": 22384
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.62640099626401e-05,
+      "loss": 0.1547,
+      "step": 22385
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6251556662515564e-05,
+      "loss": 0.2074,
+      "step": 22386
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.623910336239103e-05,
+      "loss": 0.0934,
+      "step": 22387
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6226650062266495e-05,
+      "loss": 0.1634,
+      "step": 22388
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.621419676214197e-05,
+      "loss": 0.0928,
+      "step": 22389
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.620174346201743e-05,
+      "loss": 0.0836,
+      "step": 22390
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.618929016189289e-05,
+      "loss": 0.2381,
+      "step": 22391
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6176836861768366e-05,
+      "loss": 0.1187,
+      "step": 22392
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.616438356164383e-05,
+      "loss": 0.0906,
+      "step": 22393
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.61519302615193e-05,
+      "loss": 0.0801,
+      "step": 22394
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6139476961394764e-05,
+      "loss": 0.0755,
+      "step": 22395
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.6127023661270237e-05,
+      "loss": 0.1597,
+      "step": 22396
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.61145703611457e-05,
+      "loss": 0.1838,
+      "step": 22397
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 4.610211706102116e-05,
+      "loss": 0.2123,
+      "step": 22398
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.608966376089663e-05,
+      "loss": 0.2154,
+      "step": 22399
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.60772104607721e-05,
+      "loss": 0.1846,
+      "step": 22400
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.6064757160647566e-05,
+      "loss": 0.0942,
+      "step": 22401
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.605230386052303e-05,
+      "loss": 0.1006,
+      "step": 22402
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.6039850560398505e-05,
+      "loss": 0.1438,
+      "step": 22403
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.602739726027397e-05,
+      "loss": 0.1892,
+      "step": 22404
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.601494396014944e-05,
+      "loss": 0.1169,
+      "step": 22405
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.6002490660024896e-05,
+      "loss": 0.0979,
+      "step": 22406
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.599003735990037e-05,
+      "loss": 0.1255,
+      "step": 22407
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.5977584059775835e-05,
+      "loss": 0.0974,
+      "step": 22408
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.59651307596513e-05,
+      "loss": 0.1542,
+      "step": 22409
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.595267745952677e-05,
+      "loss": 0.1271,
+      "step": 22410
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.594022415940224e-05,
+      "loss": 0.1434,
+      "step": 22411
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.5927770859277705e-05,
+      "loss": 0.1542,
+      "step": 22412
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.591531755915318e-05,
+      "loss": 0.089,
+      "step": 22413
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.590286425902864e-05,
+      "loss": 0.1411,
+      "step": 22414
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.58904109589041e-05,
+      "loss": 0.0965,
+      "step": 22415
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.587795765877957e-05,
+      "loss": 0.2054,
+      "step": 22416
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.586550435865504e-05,
+      "loss": 0.1528,
+      "step": 22417
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.585305105853051e-05,
+      "loss": 0.1179,
+      "step": 22418
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.5840597758405974e-05,
+      "loss": 0.1798,
+      "step": 22419
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.5828144458281446e-05,
+      "loss": 0.1328,
+      "step": 22420
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.581569115815691e-05,
+      "loss": 0.0955,
+      "step": 22421
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.580323785803237e-05,
+      "loss": 0.1632,
+      "step": 22422
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.579078455790784e-05,
+      "loss": 0.1653,
+      "step": 22423
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 4.577833125778331e-05,
+      "loss": 0.1613,
+      "step": 22424
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5765877957658776e-05,
+      "loss": 0.176,
+      "step": 22425
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.575342465753424e-05,
+      "loss": 0.1335,
+      "step": 22426
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5740971357409715e-05,
+      "loss": 0.0899,
+      "step": 22427
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.572851805728518e-05,
+      "loss": 0.129,
+      "step": 22428
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.571606475716065e-05,
+      "loss": 0.2173,
+      "step": 22429
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5703611457036106e-05,
+      "loss": 0.122,
+      "step": 22430
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.569115815691158e-05,
+      "loss": 0.1265,
+      "step": 22431
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5678704856787044e-05,
+      "loss": 0.1176,
+      "step": 22432
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.566625155666251e-05,
+      "loss": 0.1217,
+      "step": 22433
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5653798256537976e-05,
+      "loss": 0.1479,
+      "step": 22434
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.564134495641345e-05,
+      "loss": 0.247,
+      "step": 22435
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5628891656288915e-05,
+      "loss": 0.1575,
+      "step": 22436
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.561643835616438e-05,
+      "loss": 0.1133,
+      "step": 22437
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.560398505603984e-05,
+      "loss": 0.1662,
+      "step": 22438
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.559153175591531e-05,
+      "loss": 0.1127,
+      "step": 22439
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.557907845579078e-05,
+      "loss": 0.1004,
+      "step": 22440
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5566625155666245e-05,
+      "loss": 0.115,
+      "step": 22441
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.555417185554172e-05,
+      "loss": 0.1363,
+      "step": 22442
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5541718555417183e-05,
+      "loss": 0.1195,
+      "step": 22443
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.552926525529265e-05,
+      "loss": 0.0839,
+      "step": 22444
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.551681195516811e-05,
+      "loss": 0.1302,
+      "step": 22445
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.550435865504358e-05,
+      "loss": 0.0945,
+      "step": 22446
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.549190535491905e-05,
+      "loss": 0.2246,
+      "step": 22447
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.547945205479451e-05,
+      "loss": 0.1752,
+      "step": 22448
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.5466998754669986e-05,
+      "loss": 0.1695,
+      "step": 22449
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 4.545454545454545e-05,
+      "loss": 0.1132,
+      "step": 22450
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.544209215442092e-05,
+      "loss": 0.0787,
+      "step": 22451
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.542963885429639e-05,
+      "loss": 0.0661,
+      "step": 22452
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.541718555417185e-05,
+      "loss": 0.1425,
+      "step": 22453
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5404732254047316e-05,
+      "loss": 0.1474,
+      "step": 22454
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.539227895392278e-05,
+      "loss": 0.1321,
+      "step": 22455
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5379825653798254e-05,
+      "loss": 0.1576,
+      "step": 22456
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.536737235367372e-05,
+      "loss": 0.0816,
+      "step": 22457
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5354919053549186e-05,
+      "loss": 0.1899,
+      "step": 22458
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.534246575342466e-05,
+      "loss": 0.1478,
+      "step": 22459
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5330012453300125e-05,
+      "loss": 0.1716,
+      "step": 22460
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5317559153175584e-05,
+      "loss": 0.1508,
+      "step": 22461
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.530510585305105e-05,
+      "loss": 0.1664,
+      "step": 22462
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.529265255292652e-05,
+      "loss": 0.0589,
+      "step": 22463
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.528019925280199e-05,
+      "loss": 0.1925,
+      "step": 22464
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5267745952677455e-05,
+      "loss": 0.0742,
+      "step": 22465
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.525529265255293e-05,
+      "loss": 0.1662,
+      "step": 22466
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.524283935242839e-05,
+      "loss": 0.137,
+      "step": 22467
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.523038605230386e-05,
+      "loss": 0.1031,
+      "step": 22468
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.521793275217932e-05,
+      "loss": 0.1015,
+      "step": 22469
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.520547945205479e-05,
+      "loss": 0.108,
+      "step": 22470
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.519302615193026e-05,
+      "loss": 0.1045,
+      "step": 22471
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.518057285180572e-05,
+      "loss": 0.2027,
+      "step": 22472
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.516811955168119e-05,
+      "loss": 0.1827,
+      "step": 22473
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.515566625155666e-05,
+      "loss": 0.1266,
+      "step": 22474
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.514321295143213e-05,
+      "loss": 0.0639,
+      "step": 22475
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 4.5130759651307593e-05,
+      "loss": 0.1108,
+      "step": 22476
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.511830635118306e-05,
+      "loss": 0.1748,
+      "step": 22477
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.5105853051058525e-05,
+      "loss": 0.113,
+      "step": 22478
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.509339975093399e-05,
+      "loss": 0.1301,
+      "step": 22479
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.508094645080946e-05,
+      "loss": 0.1387,
+      "step": 22480
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.506849315068493e-05,
+      "loss": 0.1212,
+      "step": 22481
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.5056039850560396e-05,
+      "loss": 0.1375,
+      "step": 22482
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.504358655043586e-05,
+      "loss": 0.0756,
+      "step": 22483
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.5031133250311335e-05,
+      "loss": 0.1267,
+      "step": 22484
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.5018679950186794e-05,
+      "loss": 0.1954,
+      "step": 22485
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.500622665006226e-05,
+      "loss": 0.1281,
+      "step": 22486
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.4993773349937726e-05,
+      "loss": 0.0821,
+      "step": 22487
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.49813200498132e-05,
+      "loss": 0.0909,
+      "step": 22488
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.4968866749688664e-05,
+      "loss": 0.101,
+      "step": 22489
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.495641344956413e-05,
+      "loss": 0.0907,
+      "step": 22490
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.49439601494396e-05,
+      "loss": 0.1146,
+      "step": 22491
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.493150684931506e-05,
+      "loss": 0.1345,
+      "step": 22492
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.491905354919053e-05,
+      "loss": 0.104,
+      "step": 22493
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.4906600249065994e-05,
+      "loss": 0.1538,
+      "step": 22494
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.489414694894147e-05,
+      "loss": 0.1135,
+      "step": 22495
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.488169364881693e-05,
+      "loss": 0.0784,
+      "step": 22496
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.48692403486924e-05,
+      "loss": 0.3377,
+      "step": 22497
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.485678704856787e-05,
+      "loss": 0.1512,
+      "step": 22498
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.484433374844334e-05,
+      "loss": 0.1194,
+      "step": 22499
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.4831880448318796e-05,
+      "loss": 0.1123,
+      "step": 22500
+    },
+    {
+      "epoch": 8.62,
+      "eval_cer": 0.10086427219696822,
+      "eval_loss": 0.40251481533050537,
+      "eval_runtime": 47.4768,
+      "eval_samples_per_second": 36.923,
+      "eval_steps_per_second": 4.634,
+      "eval_wer": 0.16027062497171563,
+      "step": 22500
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.481942714819426e-05,
+      "loss": 0.0826,
+      "step": 22501
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 4.4806973848069735e-05,
+      "loss": 0.1274,
+      "step": 22502
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.47945205479452e-05,
+      "loss": 0.1935,
+      "step": 22503
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.478206724782067e-05,
+      "loss": 0.2453,
+      "step": 22504
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.476961394769614e-05,
+      "loss": 0.1574,
+      "step": 22505
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.4757160647571606e-05,
+      "loss": 0.1536,
+      "step": 22506
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.474470734744707e-05,
+      "loss": 0.0828,
+      "step": 22507
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.473225404732253e-05,
+      "loss": 0.086,
+      "step": 22508
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.4719800747198004e-05,
+      "loss": 0.1249,
+      "step": 22509
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.470734744707347e-05,
+      "loss": 0.2116,
+      "step": 22510
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.4694894146948935e-05,
+      "loss": 0.1102,
+      "step": 22511
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.468244084682441e-05,
+      "loss": 0.2133,
+      "step": 22512
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.4669987546699874e-05,
+      "loss": 0.1686,
+      "step": 22513
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.465753424657534e-05,
+      "loss": 0.085,
+      "step": 22514
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.4645080946450806e-05,
+      "loss": 0.1254,
+      "step": 22515
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.463262764632627e-05,
+      "loss": 0.1927,
+      "step": 22516
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.462017434620174e-05,
+      "loss": 0.2302,
+      "step": 22517
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.4607721046077204e-05,
+      "loss": 0.1066,
+      "step": 22518
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.459526774595267e-05,
+      "loss": 0.0831,
+      "step": 22519
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.458281444582814e-05,
+      "loss": 0.1333,
+      "step": 22520
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.457036114570361e-05,
+      "loss": 0.1206,
+      "step": 22521
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.4557907845579074e-05,
+      "loss": 0.1501,
+      "step": 22522
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.454545454545455e-05,
+      "loss": 0.1828,
+      "step": 22523
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.4533001245330006e-05,
+      "loss": 0.2917,
+      "step": 22524
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.452054794520547e-05,
+      "loss": 0.1007,
+      "step": 22525
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.450809464508094e-05,
+      "loss": 0.0764,
+      "step": 22526
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.449564134495641e-05,
+      "loss": 0.0836,
+      "step": 22527
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 4.448318804483188e-05,
+      "loss": 0.0729,
+      "step": 22528
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.447073474470734e-05,
+      "loss": 0.2292,
+      "step": 22529
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.4458281444582815e-05,
+      "loss": 0.1146,
+      "step": 22530
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.444582814445828e-05,
+      "loss": 0.1194,
+      "step": 22531
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.443337484433374e-05,
+      "loss": 0.1159,
+      "step": 22532
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.4420921544209207e-05,
+      "loss": 0.1426,
+      "step": 22533
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.440846824408468e-05,
+      "loss": 0.0866,
+      "step": 22534
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.4396014943960145e-05,
+      "loss": 0.1489,
+      "step": 22535
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.438356164383561e-05,
+      "loss": 0.17,
+      "step": 22536
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.4371108343711084e-05,
+      "loss": 0.1546,
+      "step": 22537
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.435865504358655e-05,
+      "loss": 0.1404,
+      "step": 22538
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.4346201743462016e-05,
+      "loss": 0.1525,
+      "step": 22539
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.4333748443337475e-05,
+      "loss": 0.0769,
+      "step": 22540
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.432129514321295e-05,
+      "loss": 0.1688,
+      "step": 22541
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.4308841843088414e-05,
+      "loss": 0.1586,
+      "step": 22542
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.429638854296388e-05,
+      "loss": 0.1299,
+      "step": 22543
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.428393524283935e-05,
+      "loss": 0.1147,
+      "step": 22544
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.427148194271482e-05,
+      "loss": 0.1374,
+      "step": 22545
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.4259028642590284e-05,
+      "loss": 0.1099,
+      "step": 22546
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.424657534246574e-05,
+      "loss": 0.2348,
+      "step": 22547
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.4234122042341216e-05,
+      "loss": 0.1773,
+      "step": 22548
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.422166874221668e-05,
+      "loss": 0.1003,
+      "step": 22549
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.420921544209215e-05,
+      "loss": 0.0955,
+      "step": 22550
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.419676214196762e-05,
+      "loss": 0.1224,
+      "step": 22551
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.4184308841843087e-05,
+      "loss": 0.1354,
+      "step": 22552
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.417185554171855e-05,
+      "loss": 0.0828,
+      "step": 22553
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 4.415940224159402e-05,
+      "loss": 0.2432,
+      "step": 22554
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.4146948941469484e-05,
+      "loss": 0.1129,
+      "step": 22555
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.413449564134495e-05,
+      "loss": 0.1432,
+      "step": 22556
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.4122042341220416e-05,
+      "loss": 0.1519,
+      "step": 22557
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.410958904109589e-05,
+      "loss": 0.1558,
+      "step": 22558
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.4097135740971355e-05,
+      "loss": 0.221,
+      "step": 22559
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.408468244084682e-05,
+      "loss": 0.215,
+      "step": 22560
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.407222914072229e-05,
+      "loss": 0.1639,
+      "step": 22561
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.405977584059776e-05,
+      "loss": 0.0635,
+      "step": 22562
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.404732254047322e-05,
+      "loss": 0.1201,
+      "step": 22563
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.4034869240348685e-05,
+      "loss": 0.1495,
+      "step": 22564
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.402241594022415e-05,
+      "loss": 0.1469,
+      "step": 22565
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.4009962640099623e-05,
+      "loss": 0.1525,
+      "step": 22566
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.399750933997509e-05,
+      "loss": 0.1597,
+      "step": 22567
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.3985056039850555e-05,
+      "loss": 0.1194,
+      "step": 22568
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.397260273972603e-05,
+      "loss": 0.1052,
+      "step": 22569
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.3960149439601494e-05,
+      "loss": 0.0957,
+      "step": 22570
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.394769613947695e-05,
+      "loss": 0.0511,
+      "step": 22571
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.393524283935242e-05,
+      "loss": 0.1637,
+      "step": 22572
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.392278953922789e-05,
+      "loss": 0.1704,
+      "step": 22573
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.391033623910336e-05,
+      "loss": 0.1541,
+      "step": 22574
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.3897882938978824e-05,
+      "loss": 0.1106,
+      "step": 22575
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.3885429638854296e-05,
+      "loss": 0.145,
+      "step": 22576
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.387297633872976e-05,
+      "loss": 0.1212,
+      "step": 22577
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.386052303860523e-05,
+      "loss": 0.2041,
+      "step": 22578
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.384806973848069e-05,
+      "loss": 0.2379,
+      "step": 22579
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 4.383561643835616e-05,
+      "loss": 0.1478,
+      "step": 22580
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.3823163138231626e-05,
+      "loss": 0.132,
+      "step": 22581
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.381070983810709e-05,
+      "loss": 0.1768,
+      "step": 22582
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.3798256537982565e-05,
+      "loss": 0.1383,
+      "step": 22583
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.378580323785803e-05,
+      "loss": 0.0783,
+      "step": 22584
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.37733499377335e-05,
+      "loss": 0.2038,
+      "step": 22585
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.376089663760897e-05,
+      "loss": 0.1036,
+      "step": 22586
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.374844333748443e-05,
+      "loss": 0.0881,
+      "step": 22587
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.3735990037359895e-05,
+      "loss": 0.0734,
+      "step": 22588
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.372353673723536e-05,
+      "loss": 0.1201,
+      "step": 22589
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.371108343711083e-05,
+      "loss": 0.1804,
+      "step": 22590
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.36986301369863e-05,
+      "loss": 0.1217,
+      "step": 22591
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.3686176836861765e-05,
+      "loss": 0.129,
+      "step": 22592
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.367372353673724e-05,
+      "loss": 0.2088,
+      "step": 22593
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.3661270236612704e-05,
+      "loss": 0.2047,
+      "step": 22594
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.364881693648816e-05,
+      "loss": 0.1621,
+      "step": 22595
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.363636363636363e-05,
+      "loss": 0.1392,
+      "step": 22596
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.36239103362391e-05,
+      "loss": 0.1745,
+      "step": 22597
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.361145703611457e-05,
+      "loss": 0.1906,
+      "step": 22598
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.3599003735990033e-05,
+      "loss": 0.1248,
+      "step": 22599
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.35865504358655e-05,
+      "loss": 0.1351,
+      "step": 22600
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.357409713574097e-05,
+      "loss": 0.1391,
+      "step": 22601
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.356164383561643e-05,
+      "loss": 0.119,
+      "step": 22602
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.35491905354919e-05,
+      "loss": 0.1371,
+      "step": 22603
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.353673723536737e-05,
+      "loss": 0.1652,
+      "step": 22604
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.3524283935242836e-05,
+      "loss": 0.1553,
+      "step": 22605
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 4.35118306351183e-05,
+      "loss": 0.1231,
+      "step": 22606
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.349937733499377e-05,
+      "loss": 0.1219,
+      "step": 22607
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.348692403486924e-05,
+      "loss": 0.1516,
+      "step": 22608
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.3474470734744706e-05,
+      "loss": 0.1785,
+      "step": 22609
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.3462017434620166e-05,
+      "loss": 0.2925,
+      "step": 22610
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.344956413449563e-05,
+      "loss": 0.2026,
+      "step": 22611
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.3437110834371104e-05,
+      "loss": 0.1841,
+      "step": 22612
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.342465753424657e-05,
+      "loss": 0.1155,
+      "step": 22613
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.3412204234122036e-05,
+      "loss": 0.1028,
+      "step": 22614
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.339975093399751e-05,
+      "loss": 0.1018,
+      "step": 22615
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.3387297633872975e-05,
+      "loss": 0.1438,
+      "step": 22616
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.337484433374844e-05,
+      "loss": 0.1354,
+      "step": 22617
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.33623910336239e-05,
+      "loss": 0.1504,
+      "step": 22618
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.334993773349937e-05,
+      "loss": 0.144,
+      "step": 22619
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.333748443337484e-05,
+      "loss": 0.0847,
+      "step": 22620
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.3325031133250305e-05,
+      "loss": 0.1085,
+      "step": 22621
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.331257783312578e-05,
+      "loss": 0.1336,
+      "step": 22622
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.330012453300124e-05,
+      "loss": 0.1261,
+      "step": 22623
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.328767123287671e-05,
+      "loss": 0.1092,
+      "step": 22624
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.327521793275218e-05,
+      "loss": 0.1722,
+      "step": 22625
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.326276463262764e-05,
+      "loss": 0.1135,
+      "step": 22626
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.325031133250311e-05,
+      "loss": 0.1048,
+      "step": 22627
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.323785803237857e-05,
+      "loss": 0.1379,
+      "step": 22628
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.3225404732254046e-05,
+      "loss": 0.2058,
+      "step": 22629
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.321295143212951e-05,
+      "loss": 0.1776,
+      "step": 22630
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.320049813200498e-05,
+      "loss": 0.1003,
+      "step": 22631
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.318804483188045e-05,
+      "loss": 0.1043,
+      "step": 22632
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 4.3175591531755916e-05,
+      "loss": 0.0957,
+      "step": 22633
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.3163138231631375e-05,
+      "loss": 0.0992,
+      "step": 22634
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.315068493150684e-05,
+      "loss": 0.1895,
+      "step": 22635
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.3138231631382314e-05,
+      "loss": 0.112,
+      "step": 22636
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.312577833125778e-05,
+      "loss": 0.1259,
+      "step": 22637
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.3113325031133246e-05,
+      "loss": 0.128,
+      "step": 22638
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.310087173100872e-05,
+      "loss": 0.1769,
+      "step": 22639
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.3088418430884185e-05,
+      "loss": 0.2753,
+      "step": 22640
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.307596513075965e-05,
+      "loss": 0.1623,
+      "step": 22641
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.306351183063511e-05,
+      "loss": 0.1487,
+      "step": 22642
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.305105853051058e-05,
+      "loss": 0.114,
+      "step": 22643
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.303860523038605e-05,
+      "loss": 0.1005,
+      "step": 22644
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.3026151930261514e-05,
+      "loss": 0.1027,
+      "step": 22645
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.301369863013698e-05,
+      "loss": 0.1721,
+      "step": 22646
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.300124533001245e-05,
+      "loss": 0.1806,
+      "step": 22647
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.298879202988792e-05,
+      "loss": 0.1276,
+      "step": 22648
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.2976338729763385e-05,
+      "loss": 0.1619,
+      "step": 22649
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.2963885429638844e-05,
+      "loss": 0.135,
+      "step": 22650
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.295143212951432e-05,
+      "loss": 0.1143,
+      "step": 22651
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.293897882938978e-05,
+      "loss": 0.0983,
+      "step": 22652
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.292652552926525e-05,
+      "loss": 0.1489,
+      "step": 22653
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.291407222914072e-05,
+      "loss": 0.1711,
+      "step": 22654
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.290161892901619e-05,
+      "loss": 0.1885,
+      "step": 22655
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.288916562889165e-05,
+      "loss": 0.116,
+      "step": 22656
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.287671232876711e-05,
+      "loss": 0.1455,
+      "step": 22657
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.2864259028642585e-05,
+      "loss": 0.0901,
+      "step": 22658
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 4.285180572851805e-05,
+      "loss": 0.1465,
+      "step": 22659
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.283935242839352e-05,
+      "loss": 0.1347,
+      "step": 22660
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.282689912826899e-05,
+      "loss": 0.1103,
+      "step": 22661
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.2814445828144456e-05,
+      "loss": 0.1345,
+      "step": 22662
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.280199252801992e-05,
+      "loss": 0.1992,
+      "step": 22663
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.2789539227895394e-05,
+      "loss": 0.1159,
+      "step": 22664
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.2777085927770854e-05,
+      "loss": 0.09,
+      "step": 22665
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.276463262764632e-05,
+      "loss": 0.181,
+      "step": 22666
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.2752179327521785e-05,
+      "loss": 0.1889,
+      "step": 22667
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.273972602739726e-05,
+      "loss": 0.0837,
+      "step": 22668
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.2727272727272724e-05,
+      "loss": 0.1905,
+      "step": 22669
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.271481942714819e-05,
+      "loss": 0.1657,
+      "step": 22670
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.270236612702366e-05,
+      "loss": 0.0789,
+      "step": 22671
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.268991282689913e-05,
+      "loss": 0.1227,
+      "step": 22672
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.267745952677459e-05,
+      "loss": 0.1912,
+      "step": 22673
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.2665006226650054e-05,
+      "loss": 0.2237,
+      "step": 22674
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.2652552926525527e-05,
+      "loss": 0.1146,
+      "step": 22675
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.264009962640099e-05,
+      "loss": 0.1298,
+      "step": 22676
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.262764632627646e-05,
+      "loss": 0.1091,
+      "step": 22677
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.261519302615193e-05,
+      "loss": 0.0713,
+      "step": 22678
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.26027397260274e-05,
+      "loss": 0.1609,
+      "step": 22679
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.259028642590286e-05,
+      "loss": 0.1162,
+      "step": 22680
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.257783312577832e-05,
+      "loss": 0.119,
+      "step": 22681
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.2565379825653795e-05,
+      "loss": 0.111,
+      "step": 22682
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.255292652552926e-05,
+      "loss": 0.0889,
+      "step": 22683
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.254047322540473e-05,
+      "loss": 0.1362,
+      "step": 22684
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 4.252801992528019e-05,
+      "loss": 0.185,
+      "step": 22685
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.2515566625155666e-05,
+      "loss": 0.1351,
+      "step": 22686
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.250311332503113e-05,
+      "loss": 0.1258,
+      "step": 22687
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.24906600249066e-05,
+      "loss": 0.1163,
+      "step": 22688
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.247820672478206e-05,
+      "loss": 0.1472,
+      "step": 22689
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.246575342465753e-05,
+      "loss": 0.1103,
+      "step": 22690
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.2453300124532995e-05,
+      "loss": 0.139,
+      "step": 22691
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.244084682440846e-05,
+      "loss": 0.1819,
+      "step": 22692
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.2428393524283934e-05,
+      "loss": 0.1347,
+      "step": 22693
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.24159402241594e-05,
+      "loss": 0.1639,
+      "step": 22694
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.2403486924034866e-05,
+      "loss": 0.1393,
+      "step": 22695
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.239103362391034e-05,
+      "loss": 0.1028,
+      "step": 22696
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.23785803237858e-05,
+      "loss": 0.18,
+      "step": 22697
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.2366127023661264e-05,
+      "loss": 0.183,
+      "step": 22698
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.235367372353673e-05,
+      "loss": 0.1104,
+      "step": 22699
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.23412204234122e-05,
+      "loss": 0.1997,
+      "step": 22700
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.232876712328767e-05,
+      "loss": 0.0776,
+      "step": 22701
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.2316313823163134e-05,
+      "loss": 0.1464,
+      "step": 22702
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.230386052303861e-05,
+      "loss": 0.1118,
+      "step": 22703
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.229140722291407e-05,
+      "loss": 0.1629,
+      "step": 22704
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.227895392278953e-05,
+      "loss": 0.1246,
+      "step": 22705
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.2266500622665e-05,
+      "loss": 0.196,
+      "step": 22706
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.225404732254047e-05,
+      "loss": 0.131,
+      "step": 22707
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.224159402241594e-05,
+      "loss": 0.0987,
+      "step": 22708
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.22291407222914e-05,
+      "loss": 0.0858,
+      "step": 22709
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.2216687422166875e-05,
+      "loss": 0.2056,
+      "step": 22710
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 4.220423412204234e-05,
+      "loss": 0.2954,
+      "step": 22711
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.21917808219178e-05,
+      "loss": 0.1428,
+      "step": 22712
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.2179327521793266e-05,
+      "loss": 0.1622,
+      "step": 22713
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.216687422166874e-05,
+      "loss": 0.1345,
+      "step": 22714
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.2154420921544205e-05,
+      "loss": 0.0709,
+      "step": 22715
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.214196762141967e-05,
+      "loss": 0.113,
+      "step": 22716
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.2129514321295144e-05,
+      "loss": 0.1438,
+      "step": 22717
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.211706102117061e-05,
+      "loss": 0.1138,
+      "step": 22718
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.2104607721046076e-05,
+      "loss": 0.1031,
+      "step": 22719
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.2092154420921535e-05,
+      "loss": 0.0992,
+      "step": 22720
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.207970112079701e-05,
+      "loss": 0.1021,
+      "step": 22721
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.2067247820672473e-05,
+      "loss": 0.1509,
+      "step": 22722
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.205479452054794e-05,
+      "loss": 0.1361,
+      "step": 22723
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.204234122042341e-05,
+      "loss": 0.1416,
+      "step": 22724
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.202988792029888e-05,
+      "loss": 0.1755,
+      "step": 22725
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.2017434620174344e-05,
+      "loss": 0.1678,
+      "step": 22726
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.200498132004981e-05,
+      "loss": 0.1796,
+      "step": 22727
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.1992528019925276e-05,
+      "loss": 0.1481,
+      "step": 22728
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.198007471980074e-05,
+      "loss": 0.1743,
+      "step": 22729
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.196762141967621e-05,
+      "loss": 0.1125,
+      "step": 22730
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.1955168119551674e-05,
+      "loss": 0.2119,
+      "step": 22731
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.1942714819427146e-05,
+      "loss": 0.1577,
+      "step": 22732
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.193026151930261e-05,
+      "loss": 0.0767,
+      "step": 22733
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.191780821917808e-05,
+      "loss": 0.1108,
+      "step": 22734
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.190535491905355e-05,
+      "loss": 0.2153,
+      "step": 22735
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.189290161892901e-05,
+      "loss": 0.1707,
+      "step": 22736
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 4.1880448318804476e-05,
+      "loss": 0.1808,
+      "step": 22737
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.186799501867994e-05,
+      "loss": 0.1605,
+      "step": 22738
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.1855541718555415e-05,
+      "loss": 0.075,
+      "step": 22739
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.184308841843088e-05,
+      "loss": 0.14,
+      "step": 22740
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.183063511830635e-05,
+      "loss": 0.2478,
+      "step": 22741
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.181818181818182e-05,
+      "loss": 0.183,
+      "step": 22742
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.1805728518057285e-05,
+      "loss": 0.0935,
+      "step": 22743
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.1793275217932745e-05,
+      "loss": 0.0767,
+      "step": 22744
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.178082191780821e-05,
+      "loss": 0.078,
+      "step": 22745
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.176836861768368e-05,
+      "loss": 0.0954,
+      "step": 22746
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.175591531755915e-05,
+      "loss": 0.1281,
+      "step": 22747
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.1743462017434615e-05,
+      "loss": 0.2655,
+      "step": 22748
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.173100871731009e-05,
+      "loss": 0.1183,
+      "step": 22749
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.1718555417185554e-05,
+      "loss": 0.1645,
+      "step": 22750
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.170610211706102e-05,
+      "loss": 0.1162,
+      "step": 22751
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.169364881693648e-05,
+      "loss": 0.1043,
+      "step": 22752
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.168119551681195e-05,
+      "loss": 0.089,
+      "step": 22753
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.166874221668742e-05,
+      "loss": 0.1754,
+      "step": 22754
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.1656288916562884e-05,
+      "loss": 0.168,
+      "step": 22755
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.1643835616438356e-05,
+      "loss": 0.102,
+      "step": 22756
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.163138231631382e-05,
+      "loss": 0.0958,
+      "step": 22757
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.161892901618929e-05,
+      "loss": 0.0689,
+      "step": 22758
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.160647571606475e-05,
+      "loss": 0.0912,
+      "step": 22759
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.159402241594022e-05,
+      "loss": 0.2016,
+      "step": 22760
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.1581569115815686e-05,
+      "loss": 0.1694,
+      "step": 22761
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.156911581569115e-05,
+      "loss": 0.2214,
+      "step": 22762
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 4.1556662515566625e-05,
+      "loss": 0.1547,
+      "step": 22763
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.154420921544209e-05,
+      "loss": 0.089,
+      "step": 22764
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.1531755915317557e-05,
+      "loss": 0.0985,
+      "step": 22765
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.151930261519302e-05,
+      "loss": 0.1392,
+      "step": 22766
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.150684931506849e-05,
+      "loss": 0.1078,
+      "step": 22767
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.1494396014943954e-05,
+      "loss": 0.1922,
+      "step": 22768
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.148194271481942e-05,
+      "loss": 0.1708,
+      "step": 22769
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.146948941469489e-05,
+      "loss": 0.1743,
+      "step": 22770
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.145703611457036e-05,
+      "loss": 0.1204,
+      "step": 22771
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.1444582814445825e-05,
+      "loss": 0.2156,
+      "step": 22772
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.143212951432129e-05,
+      "loss": 0.1876,
+      "step": 22773
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.1419676214196764e-05,
+      "loss": 0.119,
+      "step": 22774
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.140722291407222e-05,
+      "loss": 0.111,
+      "step": 22775
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.139476961394769e-05,
+      "loss": 0.0634,
+      "step": 22776
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.1382316313823155e-05,
+      "loss": 0.1732,
+      "step": 22777
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.136986301369863e-05,
+      "loss": 0.057,
+      "step": 22778
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.135740971357409e-05,
+      "loss": 0.229,
+      "step": 22779
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.134495641344956e-05,
+      "loss": 0.1794,
+      "step": 22780
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.133250311332503e-05,
+      "loss": 0.1164,
+      "step": 22781
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.13200498132005e-05,
+      "loss": 0.0952,
+      "step": 22782
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.130759651307596e-05,
+      "loss": 0.0984,
+      "step": 22783
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.129514321295142e-05,
+      "loss": 0.1854,
+      "step": 22784
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.1282689912826896e-05,
+      "loss": 0.1165,
+      "step": 22785
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.127023661270236e-05,
+      "loss": 0.1376,
+      "step": 22786
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.125778331257783e-05,
+      "loss": 0.0889,
+      "step": 22787
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.12453300124533e-05,
+      "loss": 0.099,
+      "step": 22788
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 4.1232876712328766e-05,
+      "loss": 0.0672,
+      "step": 22789
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.122042341220423e-05,
+      "loss": 0.1117,
+      "step": 22790
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.120797011207969e-05,
+      "loss": 0.1581,
+      "step": 22791
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.1195516811955164e-05,
+      "loss": 0.1449,
+      "step": 22792
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.118306351183063e-05,
+      "loss": 0.1339,
+      "step": 22793
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.1170610211706096e-05,
+      "loss": 0.0965,
+      "step": 22794
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.115815691158157e-05,
+      "loss": 0.1154,
+      "step": 22795
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.1145703611457035e-05,
+      "loss": 0.1425,
+      "step": 22796
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.11332503113325e-05,
+      "loss": 0.1041,
+      "step": 22797
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.112079701120797e-05,
+      "loss": 0.1802,
+      "step": 22798
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.110834371108343e-05,
+      "loss": 0.1226,
+      "step": 22799
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.10958904109589e-05,
+      "loss": 0.1047,
+      "step": 22800
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.1083437110834364e-05,
+      "loss": 0.0969,
+      "step": 22801
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.107098381070984e-05,
+      "loss": 0.0712,
+      "step": 22802
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.10585305105853e-05,
+      "loss": 0.1281,
+      "step": 22803
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.104607721046077e-05,
+      "loss": 0.3014,
+      "step": 22804
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.103362391033624e-05,
+      "loss": 0.3049,
+      "step": 22805
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.102117061021171e-05,
+      "loss": 0.1666,
+      "step": 22806
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.100871731008717e-05,
+      "loss": 0.0882,
+      "step": 22807
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.099626400996263e-05,
+      "loss": 0.1184,
+      "step": 22808
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.0983810709838106e-05,
+      "loss": 0.1079,
+      "step": 22809
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.097135740971357e-05,
+      "loss": 0.1216,
+      "step": 22810
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.095890410958904e-05,
+      "loss": 0.1368,
+      "step": 22811
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.09464508094645e-05,
+      "loss": 0.1048,
+      "step": 22812
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.0933997509339976e-05,
+      "loss": 0.0995,
+      "step": 22813
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.0921544209215435e-05,
+      "loss": 0.1128,
+      "step": 22814
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 4.09090909090909e-05,
+      "loss": 0.0865,
+      "step": 22815
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.0896637608966374e-05,
+      "loss": 0.1837,
+      "step": 22816
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.088418430884184e-05,
+      "loss": 0.1305,
+      "step": 22817
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.0871731008717306e-05,
+      "loss": 0.1318,
+      "step": 22818
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.085927770859277e-05,
+      "loss": 0.0801,
+      "step": 22819
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.0846824408468244e-05,
+      "loss": 0.0787,
+      "step": 22820
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.083437110834371e-05,
+      "loss": 0.1105,
+      "step": 22821
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.082191780821917e-05,
+      "loss": 0.1108,
+      "step": 22822
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.0809464508094636e-05,
+      "loss": 0.1836,
+      "step": 22823
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.079701120797011e-05,
+      "loss": 0.1076,
+      "step": 22824
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.0784557907845574e-05,
+      "loss": 0.1134,
+      "step": 22825
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.077210460772104e-05,
+      "loss": 0.108,
+      "step": 22826
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.075965130759651e-05,
+      "loss": 0.0935,
+      "step": 22827
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.074719800747198e-05,
+      "loss": 0.0732,
+      "step": 22828
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.0734744707347445e-05,
+      "loss": 0.2331,
+      "step": 22829
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.0722291407222904e-05,
+      "loss": 0.0852,
+      "step": 22830
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.070983810709838e-05,
+      "loss": 0.1372,
+      "step": 22831
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.069738480697384e-05,
+      "loss": 0.1009,
+      "step": 22832
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.068493150684931e-05,
+      "loss": 0.0937,
+      "step": 22833
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.067247820672478e-05,
+      "loss": 0.1779,
+      "step": 22834
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.066002490660025e-05,
+      "loss": 0.2169,
+      "step": 22835
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.064757160647571e-05,
+      "loss": 0.1043,
+      "step": 22836
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.0635118306351186e-05,
+      "loss": 0.1854,
+      "step": 22837
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.0622665006226645e-05,
+      "loss": 0.1069,
+      "step": 22838
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.061021170610211e-05,
+      "loss": 0.1094,
+      "step": 22839
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.059775840597758e-05,
+      "loss": 0.0791,
+      "step": 22840
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 4.058530510585305e-05,
+      "loss": 0.1578,
+      "step": 22841
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.0572851805728516e-05,
+      "loss": 0.2122,
+      "step": 22842
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.056039850560398e-05,
+      "loss": 0.1428,
+      "step": 22843
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.0547945205479454e-05,
+      "loss": 0.1399,
+      "step": 22844
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.053549190535492e-05,
+      "loss": 0.0744,
+      "step": 22845
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.052303860523038e-05,
+      "loss": 0.0746,
+      "step": 22846
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.0510585305105845e-05,
+      "loss": 0.1493,
+      "step": 22847
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.049813200498132e-05,
+      "loss": 0.1702,
+      "step": 22848
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.0485678704856784e-05,
+      "loss": 0.1872,
+      "step": 22849
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.047322540473225e-05,
+      "loss": 0.1067,
+      "step": 22850
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.046077210460772e-05,
+      "loss": 0.1205,
+      "step": 22851
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.044831880448319e-05,
+      "loss": 0.1188,
+      "step": 22852
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.0435865504358655e-05,
+      "loss": 0.0697,
+      "step": 22853
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.0423412204234114e-05,
+      "loss": 0.1737,
+      "step": 22854
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.0410958904109586e-05,
+      "loss": 0.1733,
+      "step": 22855
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.039850560398505e-05,
+      "loss": 0.0913,
+      "step": 22856
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.038605230386052e-05,
+      "loss": 0.1256,
+      "step": 22857
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.0373599003735984e-05,
+      "loss": 0.0936,
+      "step": 22858
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.036114570361146e-05,
+      "loss": 0.1246,
+      "step": 22859
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.034869240348692e-05,
+      "loss": 0.2694,
+      "step": 22860
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.033623910336239e-05,
+      "loss": 0.1715,
+      "step": 22861
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.032378580323785e-05,
+      "loss": 0.1002,
+      "step": 22862
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.031133250311332e-05,
+      "loss": 0.1378,
+      "step": 22863
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.029887920298879e-05,
+      "loss": 0.0755,
+      "step": 22864
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.028642590286425e-05,
+      "loss": 0.1947,
+      "step": 22865
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.0273972602739725e-05,
+      "loss": 0.1148,
+      "step": 22866
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 4.026151930261519e-05,
+      "loss": 0.1346,
+      "step": 22867
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.024906600249066e-05,
+      "loss": 0.2038,
+      "step": 22868
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.0236612702366116e-05,
+      "loss": 0.1596,
+      "step": 22869
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.022415940224159e-05,
+      "loss": 0.0749,
+      "step": 22870
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.0211706102117055e-05,
+      "loss": 0.0873,
+      "step": 22871
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.019925280199252e-05,
+      "loss": 0.1234,
+      "step": 22872
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.0186799501867994e-05,
+      "loss": 0.1159,
+      "step": 22873
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.017434620174346e-05,
+      "loss": 0.1403,
+      "step": 22874
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.0161892901618926e-05,
+      "loss": 0.1061,
+      "step": 22875
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.01494396014944e-05,
+      "loss": 0.0863,
+      "step": 22876
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.013698630136986e-05,
+      "loss": 0.0711,
+      "step": 22877
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.0124533001245323e-05,
+      "loss": 0.2138,
+      "step": 22878
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.011207970112079e-05,
+      "loss": 0.1437,
+      "step": 22879
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.009962640099626e-05,
+      "loss": 0.3046,
+      "step": 22880
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.008717310087173e-05,
+      "loss": 0.1601,
+      "step": 22881
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.0074719800747194e-05,
+      "loss": 0.2268,
+      "step": 22882
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.006226650062267e-05,
+      "loss": 0.1535,
+      "step": 22883
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.004981320049813e-05,
+      "loss": 0.0916,
+      "step": 22884
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.003735990037359e-05,
+      "loss": 0.1801,
+      "step": 22885
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.002490660024906e-05,
+      "loss": 0.2055,
+      "step": 22886
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 4.001245330012453e-05,
+      "loss": 0.2078,
+      "step": 22887
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.9999999999999996e-05,
+      "loss": 0.1205,
+      "step": 22888
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.998754669987546e-05,
+      "loss": 0.1266,
+      "step": 22889
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.9975093399750935e-05,
+      "loss": 0.1452,
+      "step": 22890
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.99626400996264e-05,
+      "loss": 0.183,
+      "step": 22891
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.995018679950187e-05,
+      "loss": 0.1795,
+      "step": 22892
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.9937733499377326e-05,
+      "loss": 0.1274,
+      "step": 22893
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.99252801992528e-05,
+      "loss": 0.0928,
+      "step": 22894
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.9912826899128265e-05,
+      "loss": 0.1422,
+      "step": 22895
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.990037359900373e-05,
+      "loss": 0.1124,
+      "step": 22896
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.98879202988792e-05,
+      "loss": 0.1688,
+      "step": 22897
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.987546699875467e-05,
+      "loss": 0.1753,
+      "step": 22898
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.9863013698630135e-05,
+      "loss": 0.1489,
+      "step": 22899
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.98505603985056e-05,
+      "loss": 0.0967,
+      "step": 22900
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.983810709838107e-05,
+      "loss": 0.1465,
+      "step": 22901
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.982565379825653e-05,
+      "loss": 0.0971,
+      "step": 22902
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.9813200498132e-05,
+      "loss": 0.1684,
+      "step": 22903
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.9800747198007465e-05,
+      "loss": 0.172,
+      "step": 22904
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.978829389788294e-05,
+      "loss": 0.1397,
+      "step": 22905
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.9775840597758404e-05,
+      "loss": 0.0825,
+      "step": 22906
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.976338729763387e-05,
+      "loss": 0.1066,
+      "step": 22907
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.975093399750934e-05,
+      "loss": 0.1133,
+      "step": 22908
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.97384806973848e-05,
+      "loss": 0.0946,
+      "step": 22909
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.972602739726027e-05,
+      "loss": 0.1915,
+      "step": 22910
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.9713574097135734e-05,
+      "loss": 0.1361,
+      "step": 22911
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.9701120797011206e-05,
+      "loss": 0.1521,
+      "step": 22912
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.968866749688667e-05,
+      "loss": 0.1197,
+      "step": 22913
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.967621419676214e-05,
+      "loss": 0.0827,
+      "step": 22914
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.966376089663761e-05,
+      "loss": 0.173,
+      "step": 22915
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.965130759651308e-05,
+      "loss": 0.192,
+      "step": 22916
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.9638854296388536e-05,
+      "loss": 0.2071,
+      "step": 22917
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.9626400996264e-05,
+      "loss": 0.123,
+      "step": 22918
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.9613947696139475e-05,
+      "loss": 0.1609,
+      "step": 22919
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.960149439601494e-05,
+      "loss": 0.108,
+      "step": 22920
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.9589041095890407e-05,
+      "loss": 0.1018,
+      "step": 22921
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.957658779576588e-05,
+      "loss": 0.164,
+      "step": 22922
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.9564134495641345e-05,
+      "loss": 0.1405,
+      "step": 22923
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.9551681195516804e-05,
+      "loss": 0.1114,
+      "step": 22924
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.953922789539227e-05,
+      "loss": 0.0946,
+      "step": 22925
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.952677459526774e-05,
+      "loss": 0.1085,
+      "step": 22926
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.951432129514321e-05,
+      "loss": 0.1434,
+      "step": 22927
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.9501867995018675e-05,
+      "loss": 0.168,
+      "step": 22928
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.948941469489415e-05,
+      "loss": 0.158,
+      "step": 22929
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.9476961394769614e-05,
+      "loss": 0.1536,
+      "step": 22930
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.946450809464508e-05,
+      "loss": 0.124,
+      "step": 22931
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.945205479452054e-05,
+      "loss": 0.0755,
+      "step": 22932
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.943960149439601e-05,
+      "loss": 0.1043,
+      "step": 22933
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.942714819427148e-05,
+      "loss": 0.188,
+      "step": 22934
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.941469489414694e-05,
+      "loss": 0.1796,
+      "step": 22935
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.9402241594022416e-05,
+      "loss": 0.1744,
+      "step": 22936
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.938978829389788e-05,
+      "loss": 0.1519,
+      "step": 22937
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.937733499377335e-05,
+      "loss": 0.0869,
+      "step": 22938
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.9364881693648814e-05,
+      "loss": 0.1102,
+      "step": 22939
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.935242839352428e-05,
+      "loss": 0.1293,
+      "step": 22940
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.9339975093399746e-05,
+      "loss": 0.1885,
+      "step": 22941
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.932752179327521e-05,
+      "loss": 0.195,
+      "step": 22942
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.931506849315068e-05,
+      "loss": 0.1583,
+      "step": 22943
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.930261519302615e-05,
+      "loss": 0.1437,
+      "step": 22944
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.9290161892901616e-05,
+      "loss": 0.2562,
+      "step": 22945
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.927770859277708e-05,
+      "loss": 0.0741,
+      "step": 22946
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.9265255292652555e-05,
+      "loss": 0.1415,
+      "step": 22947
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.9252801992528014e-05,
+      "loss": 0.1061,
+      "step": 22948
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.924034869240348e-05,
+      "loss": 0.1541,
+      "step": 22949
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.9227895392278946e-05,
+      "loss": 0.1024,
+      "step": 22950
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.921544209215442e-05,
+      "loss": 0.108,
+      "step": 22951
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.9202988792029885e-05,
+      "loss": 0.1177,
+      "step": 22952
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.919053549190535e-05,
+      "loss": 0.1159,
+      "step": 22953
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.9178082191780823e-05,
+      "loss": 0.2086,
+      "step": 22954
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.916562889165629e-05,
+      "loss": 0.1326,
+      "step": 22955
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.915317559153175e-05,
+      "loss": 0.1699,
+      "step": 22956
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.9140722291407214e-05,
+      "loss": 0.1457,
+      "step": 22957
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.912826899128269e-05,
+      "loss": 0.1923,
+      "step": 22958
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.911581569115815e-05,
+      "loss": 0.0652,
+      "step": 22959
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.910336239103362e-05,
+      "loss": 0.2012,
+      "step": 22960
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.909090909090909e-05,
+      "loss": 0.1328,
+      "step": 22961
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.907845579078456e-05,
+      "loss": 0.1073,
+      "step": 22962
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.9066002490660024e-05,
+      "loss": 0.2742,
+      "step": 22963
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.905354919053548e-05,
+      "loss": 0.0976,
+      "step": 22964
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.9041095890410956e-05,
+      "loss": 0.0952,
+      "step": 22965
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.902864259028642e-05,
+      "loss": 0.1213,
+      "step": 22966
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.901618929016189e-05,
+      "loss": 0.1744,
+      "step": 22967
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.900373599003736e-05,
+      "loss": 0.1698,
+      "step": 22968
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.8991282689912826e-05,
+      "loss": 0.1057,
+      "step": 22969
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.897882938978829e-05,
+      "loss": 0.1397,
+      "step": 22970
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.896637608966375e-05,
+      "loss": 0.1626,
+      "step": 22971
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.8953922789539224e-05,
+      "loss": 0.1171,
+      "step": 22972
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.894146948941469e-05,
+      "loss": 0.1436,
+      "step": 22973
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.8929016189290156e-05,
+      "loss": 0.1034,
+      "step": 22974
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.891656288916563e-05,
+      "loss": 0.0739,
+      "step": 22975
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.8904109589041095e-05,
+      "loss": 0.0697,
+      "step": 22976
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.889165628891656e-05,
+      "loss": 0.1376,
+      "step": 22977
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.8879202988792026e-05,
+      "loss": 0.1903,
+      "step": 22978
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.886674968866749e-05,
+      "loss": 0.1955,
+      "step": 22979
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.885429638854296e-05,
+      "loss": 0.1099,
+      "step": 22980
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.8841843088418424e-05,
+      "loss": 0.1224,
+      "step": 22981
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.88293897882939e-05,
+      "loss": 0.1518,
+      "step": 22982
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.881693648816936e-05,
+      "loss": 0.1299,
+      "step": 22983
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.880448318804483e-05,
+      "loss": 0.1553,
+      "step": 22984
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.8792029887920295e-05,
+      "loss": 0.1515,
+      "step": 22985
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.877957658779577e-05,
+      "loss": 0.2159,
+      "step": 22986
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.876712328767123e-05,
+      "loss": 0.129,
+      "step": 22987
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.875466998754669e-05,
+      "loss": 0.1876,
+      "step": 22988
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.874221668742216e-05,
+      "loss": 0.1399,
+      "step": 22989
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.872976338729763e-05,
+      "loss": 0.1228,
+      "step": 22990
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.87173100871731e-05,
+      "loss": 0.1827,
+      "step": 22991
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.870485678704856e-05,
+      "loss": 0.2341,
+      "step": 22992
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.8692403486924036e-05,
+      "loss": 0.1048,
+      "step": 22993
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.86799501867995e-05,
+      "loss": 0.1116,
+      "step": 22994
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.866749688667496e-05,
+      "loss": 0.1646,
+      "step": 22995
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.865504358655043e-05,
+      "loss": 0.1087,
+      "step": 22996
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.86425902864259e-05,
+      "loss": 0.1419,
+      "step": 22997
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 3.8630136986301366e-05,
+      "loss": 0.1635,
+      "step": 22998
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.861768368617683e-05,
+      "loss": 0.15,
+      "step": 22999
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.8605230386052304e-05,
+      "loss": 0.1606,
+      "step": 23000
+    },
+    {
+      "epoch": 8.82,
+      "eval_cer": 0.10044286634020719,
+      "eval_loss": 0.3962526321411133,
+      "eval_runtime": 46.8342,
+      "eval_samples_per_second": 37.43,
+      "eval_steps_per_second": 4.697,
+      "eval_wer": 0.15800787437208672,
+      "step": 23000
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.859277708592777e-05,
+      "loss": 0.1018,
+      "step": 23001
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.8580323785803236e-05,
+      "loss": 0.0966,
+      "step": 23002
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.8567870485678695e-05,
+      "loss": 0.1226,
+      "step": 23003
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.855541718555417e-05,
+      "loss": 0.1458,
+      "step": 23004
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.8542963885429634e-05,
+      "loss": 0.2307,
+      "step": 23005
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.85305105853051e-05,
+      "loss": 0.0992,
+      "step": 23006
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.851805728518057e-05,
+      "loss": 0.0521,
+      "step": 23007
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.850560398505604e-05,
+      "loss": 0.1107,
+      "step": 23008
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.8493150684931505e-05,
+      "loss": 0.0961,
+      "step": 23009
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.848069738480698e-05,
+      "loss": 0.1914,
+      "step": 23010
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.8468244084682436e-05,
+      "loss": 0.1448,
+      "step": 23011
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.84557907845579e-05,
+      "loss": 0.1323,
+      "step": 23012
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.844333748443337e-05,
+      "loss": 0.0724,
+      "step": 23013
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.843088418430884e-05,
+      "loss": 0.095,
+      "step": 23014
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.841843088418431e-05,
+      "loss": 0.1894,
+      "step": 23015
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.840597758405977e-05,
+      "loss": 0.105,
+      "step": 23016
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.8393524283935246e-05,
+      "loss": 0.2202,
+      "step": 23017
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.838107098381071e-05,
+      "loss": 0.1371,
+      "step": 23018
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.836861768368617e-05,
+      "loss": 0.0722,
+      "step": 23019
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.835616438356164e-05,
+      "loss": 0.0833,
+      "step": 23020
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.834371108343711e-05,
+      "loss": 0.2245,
+      "step": 23021
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.8331257783312575e-05,
+      "loss": 0.1632,
+      "step": 23022
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.831880448318804e-05,
+      "loss": 0.1743,
+      "step": 23023
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.830635118306351e-05,
+      "loss": 0.1217,
+      "step": 23024
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.829389788293898e-05,
+      "loss": 0.1569,
+      "step": 23025
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.828144458281444e-05,
+      "loss": 0.0937,
+      "step": 23026
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.8268991282689905e-05,
+      "loss": 0.1184,
+      "step": 23027
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.825653798256538e-05,
+      "loss": 0.1485,
+      "step": 23028
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.8244084682440844e-05,
+      "loss": 0.1794,
+      "step": 23029
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.823163138231631e-05,
+      "loss": 0.0977,
+      "step": 23030
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.8219178082191776e-05,
+      "loss": 0.2404,
+      "step": 23031
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.820672478206725e-05,
+      "loss": 0.172,
+      "step": 23032
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.8194271481942714e-05,
+      "loss": 0.0907,
+      "step": 23033
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.8181818181818174e-05,
+      "loss": 0.0863,
+      "step": 23034
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.816936488169364e-05,
+      "loss": 0.1704,
+      "step": 23035
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.815691158156911e-05,
+      "loss": 0.1185,
+      "step": 23036
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.814445828144458e-05,
+      "loss": 0.1631,
+      "step": 23037
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.8132004981320044e-05,
+      "loss": 0.0958,
+      "step": 23038
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.811955168119552e-05,
+      "loss": 0.1481,
+      "step": 23039
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.810709838107098e-05,
+      "loss": 0.1395,
+      "step": 23040
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.809464508094645e-05,
+      "loss": 0.117,
+      "step": 23041
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.808219178082191e-05,
+      "loss": 0.1377,
+      "step": 23042
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.806973848069738e-05,
+      "loss": 0.109,
+      "step": 23043
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.8057285180572847e-05,
+      "loss": 0.0709,
+      "step": 23044
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.804483188044831e-05,
+      "loss": 0.1062,
+      "step": 23045
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.8032378580323785e-05,
+      "loss": 0.1078,
+      "step": 23046
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.801992528019925e-05,
+      "loss": 0.1809,
+      "step": 23047
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.800747198007472e-05,
+      "loss": 0.1494,
+      "step": 23048
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.799501867995019e-05,
+      "loss": 0.2171,
+      "step": 23049
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 3.798256537982565e-05,
+      "loss": 0.1252,
+      "step": 23050
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.7970112079701115e-05,
+      "loss": 0.0975,
+      "step": 23051
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.795765877957658e-05,
+      "loss": 0.1026,
+      "step": 23052
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.7945205479452054e-05,
+      "loss": 0.149,
+      "step": 23053
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.793275217932752e-05,
+      "loss": 0.1894,
+      "step": 23054
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.7920298879202985e-05,
+      "loss": 0.1649,
+      "step": 23055
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.790784557907846e-05,
+      "loss": 0.1087,
+      "step": 23056
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.7895392278953924e-05,
+      "loss": 0.1795,
+      "step": 23057
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.788293897882938e-05,
+      "loss": 0.0961,
+      "step": 23058
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.787048567870485e-05,
+      "loss": 0.1336,
+      "step": 23059
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.785803237858032e-05,
+      "loss": 0.1149,
+      "step": 23060
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.784557907845579e-05,
+      "loss": 0.188,
+      "step": 23061
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.7833125778331254e-05,
+      "loss": 0.0783,
+      "step": 23062
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.7820672478206727e-05,
+      "loss": 0.1118,
+      "step": 23063
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.780821917808219e-05,
+      "loss": 0.0873,
+      "step": 23064
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.779576587795766e-05,
+      "loss": 0.1862,
+      "step": 23065
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.778331257783312e-05,
+      "loss": 0.3013,
+      "step": 23066
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.777085927770859e-05,
+      "loss": 0.1707,
+      "step": 23067
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.7758405977584056e-05,
+      "loss": 0.0976,
+      "step": 23068
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.774595267745952e-05,
+      "loss": 0.1595,
+      "step": 23069
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.773349937733499e-05,
+      "loss": 0.1048,
+      "step": 23070
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.772104607721046e-05,
+      "loss": 0.072,
+      "step": 23071
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.770859277708593e-05,
+      "loss": 0.1261,
+      "step": 23072
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.769613947696139e-05,
+      "loss": 0.1299,
+      "step": 23073
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.768368617683685e-05,
+      "loss": 0.1258,
+      "step": 23074
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.7671232876712325e-05,
+      "loss": 0.17,
+      "step": 23075
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.765877957658779e-05,
+      "loss": 0.106,
+      "step": 23076
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.764632627646326e-05,
+      "loss": 0.0871,
+      "step": 23077
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.763387297633873e-05,
+      "loss": 0.0769,
+      "step": 23078
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.7621419676214195e-05,
+      "loss": 0.2209,
+      "step": 23079
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.760896637608966e-05,
+      "loss": 0.1548,
+      "step": 23080
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.759651307596512e-05,
+      "loss": 0.0991,
+      "step": 23081
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.758405977584059e-05,
+      "loss": 0.0997,
+      "step": 23082
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.757160647571606e-05,
+      "loss": 0.1699,
+      "step": 23083
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.7559153175591525e-05,
+      "loss": 0.1916,
+      "step": 23084
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.7546699875467e-05,
+      "loss": 0.2073,
+      "step": 23085
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.7534246575342464e-05,
+      "loss": 0.1308,
+      "step": 23086
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.752179327521793e-05,
+      "loss": 0.1087,
+      "step": 23087
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.75093399750934e-05,
+      "loss": 0.156,
+      "step": 23088
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.749688667496886e-05,
+      "loss": 0.0754,
+      "step": 23089
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.7484433374844334e-05,
+      "loss": 0.1108,
+      "step": 23090
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.7471980074719793e-05,
+      "loss": 0.2626,
+      "step": 23091
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.7459526774595266e-05,
+      "loss": 0.1254,
+      "step": 23092
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.744707347447073e-05,
+      "loss": 0.1447,
+      "step": 23093
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.74346201743462e-05,
+      "loss": 0.1061,
+      "step": 23094
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.7422166874221664e-05,
+      "loss": 0.128,
+      "step": 23095
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.740971357409713e-05,
+      "loss": 0.0764,
+      "step": 23096
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.73972602739726e-05,
+      "loss": 0.1462,
+      "step": 23097
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.738480697384807e-05,
+      "loss": 0.1274,
+      "step": 23098
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.7372353673723535e-05,
+      "loss": 0.2772,
+      "step": 23099
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.7359900373599e-05,
+      "loss": 0.114,
+      "step": 23100
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.7347447073474466e-05,
+      "loss": 0.2059,
+      "step": 23101
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.733499377334994e-05,
+      "loss": 0.0814,
+      "step": 23102
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.73225404732254e-05,
+      "loss": 0.1435,
+      "step": 23103
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.731008717310087e-05,
+      "loss": 0.1312,
+      "step": 23104
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.729763387297634e-05,
+      "loss": 0.1619,
+      "step": 23105
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.72851805728518e-05,
+      "loss": 0.1424,
+      "step": 23106
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.727272727272727e-05,
+      "loss": 0.1354,
+      "step": 23107
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.7260273972602735e-05,
+      "loss": 0.1355,
+      "step": 23108
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.72478206724782e-05,
+      "loss": 0.0879,
+      "step": 23109
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.7235367372353673e-05,
+      "loss": 0.1874,
+      "step": 23110
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.722291407222914e-05,
+      "loss": 0.1858,
+      "step": 23111
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.7210460772104605e-05,
+      "loss": 0.0812,
+      "step": 23112
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.719800747198007e-05,
+      "loss": 0.1284,
+      "step": 23113
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.718555417185554e-05,
+      "loss": 0.0789,
+      "step": 23114
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.7173100871731e-05,
+      "loss": 0.1229,
+      "step": 23115
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.716064757160647e-05,
+      "loss": 0.1256,
+      "step": 23116
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.714819427148194e-05,
+      "loss": 0.0942,
+      "step": 23117
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.713574097135741e-05,
+      "loss": 0.2326,
+      "step": 23118
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.7123287671232874e-05,
+      "loss": 0.099,
+      "step": 23119
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.711083437110834e-05,
+      "loss": 0.1072,
+      "step": 23120
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.7098381070983806e-05,
+      "loss": 0.1197,
+      "step": 23121
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.708592777085928e-05,
+      "loss": 0.2334,
+      "step": 23122
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.707347447073474e-05,
+      "loss": 0.151,
+      "step": 23123
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.706102117061021e-05,
+      "loss": 0.1286,
+      "step": 23124
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.7048567870485676e-05,
+      "loss": 0.2507,
+      "step": 23125
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.703611457036114e-05,
+      "loss": 0.1882,
+      "step": 23126
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.702366127023661e-05,
+      "loss": 0.0729,
+      "step": 23127
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.7011207970112074e-05,
+      "loss": 0.1146,
+      "step": 23128
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.699875466998755e-05,
+      "loss": 0.1675,
+      "step": 23129
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.6986301369863006e-05,
+      "loss": 0.1089,
+      "step": 23130
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.697384806973848e-05,
+      "loss": 0.1211,
+      "step": 23131
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.6961394769613945e-05,
+      "loss": 0.1223,
+      "step": 23132
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.694894146948941e-05,
+      "loss": 0.0648,
+      "step": 23133
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.6936488169364876e-05,
+      "loss": 0.0831,
+      "step": 23134
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.692403486924034e-05,
+      "loss": 0.2206,
+      "step": 23135
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.6911581569115815e-05,
+      "loss": 0.2273,
+      "step": 23136
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.689912826899128e-05,
+      "loss": 0.1877,
+      "step": 23137
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.688667496886675e-05,
+      "loss": 0.0913,
+      "step": 23138
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.687422166874221e-05,
+      "loss": 0.1194,
+      "step": 23139
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.686176836861768e-05,
+      "loss": 0.0537,
+      "step": 23140
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.684931506849315e-05,
+      "loss": 0.1971,
+      "step": 23141
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.683686176836861e-05,
+      "loss": 0.1977,
+      "step": 23142
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.6824408468244084e-05,
+      "loss": 0.1861,
+      "step": 23143
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.681195516811955e-05,
+      "loss": 0.254,
+      "step": 23144
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.6799501867995015e-05,
+      "loss": 0.0642,
+      "step": 23145
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.678704856787048e-05,
+      "loss": 0.0874,
+      "step": 23146
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.677459526774595e-05,
+      "loss": 0.1907,
+      "step": 23147
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.676214196762142e-05,
+      "loss": 0.1476,
+      "step": 23148
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.6749688667496886e-05,
+      "loss": 0.121,
+      "step": 23149
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.673723536737235e-05,
+      "loss": 0.0921,
+      "step": 23150
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.672478206724782e-05,
+      "loss": 0.1193,
+      "step": 23151
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.6712328767123284e-05,
+      "loss": 0.1304,
+      "step": 23152
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.669987546699875e-05,
+      "loss": 0.1401,
+      "step": 23153
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.6687422166874216e-05,
+      "loss": 0.1586,
+      "step": 23154
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.667496886674968e-05,
+      "loss": 0.1549,
+      "step": 23155
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.6662515566625154e-05,
+      "loss": 0.131,
+      "step": 23156
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.665006226650062e-05,
+      "loss": 0.0792,
+      "step": 23157
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.6637608966376086e-05,
+      "loss": 0.1218,
+      "step": 23158
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.662515566625155e-05,
+      "loss": 0.1339,
+      "step": 23159
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.661270236612702e-05,
+      "loss": 0.151,
+      "step": 23160
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.660024906600249e-05,
+      "loss": 0.1432,
+      "step": 23161
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.658779576587795e-05,
+      "loss": 0.1527,
+      "step": 23162
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.657534246575342e-05,
+      "loss": 0.0996,
+      "step": 23163
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.656288916562889e-05,
+      "loss": 0.0957,
+      "step": 23164
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.6550435865504355e-05,
+      "loss": 0.0762,
+      "step": 23165
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.653798256537982e-05,
+      "loss": 0.2418,
+      "step": 23166
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.6525529265255287e-05,
+      "loss": 0.1559,
+      "step": 23167
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.651307596513076e-05,
+      "loss": 0.0919,
+      "step": 23168
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.6500622665006225e-05,
+      "loss": 0.1895,
+      "step": 23169
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.648816936488169e-05,
+      "loss": 0.1238,
+      "step": 23170
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.647571606475716e-05,
+      "loss": 0.1772,
+      "step": 23171
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.646326276463262e-05,
+      "loss": 0.3332,
+      "step": 23172
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.6450809464508096e-05,
+      "loss": 0.1771,
+      "step": 23173
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.6438356164383555e-05,
+      "loss": 0.2268,
+      "step": 23174
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.642590286425903e-05,
+      "loss": 0.1113,
+      "step": 23175
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.6413449564134494e-05,
+      "loss": 0.1035,
+      "step": 23176
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.640099626400996e-05,
+      "loss": 0.0831,
+      "step": 23177
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.6388542963885425e-05,
+      "loss": 0.0977,
+      "step": 23178
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.637608966376089e-05,
+      "loss": 0.1726,
+      "step": 23179
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.1666,
+      "step": 23180
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.635118306351183e-05,
+      "loss": 0.1803,
+      "step": 23181
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.6338729763387296e-05,
+      "loss": 0.1272,
+      "step": 23182
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.632627646326276e-05,
+      "loss": 0.1623,
+      "step": 23183
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.631382316313823e-05,
+      "loss": 0.0561,
+      "step": 23184
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.6301369863013694e-05,
+      "loss": 0.1752,
+      "step": 23185
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.628891656288916e-05,
+      "loss": 0.1502,
+      "step": 23186
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.627646326276463e-05,
+      "loss": 0.2058,
+      "step": 23187
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.62640099626401e-05,
+      "loss": 0.2705,
+      "step": 23188
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.6251556662515564e-05,
+      "loss": 0.0766,
+      "step": 23189
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.623910336239103e-05,
+      "loss": 0.1072,
+      "step": 23190
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.6226650062266496e-05,
+      "loss": 0.101,
+      "step": 23191
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.621419676214197e-05,
+      "loss": 0.1503,
+      "step": 23192
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.620174346201743e-05,
+      "loss": 0.1451,
+      "step": 23193
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.61892901618929e-05,
+      "loss": 0.1599,
+      "step": 23194
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.617683686176837e-05,
+      "loss": 0.0795,
+      "step": 23195
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.616438356164383e-05,
+      "loss": 0.1376,
+      "step": 23196
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.61519302615193e-05,
+      "loss": 0.1525,
+      "step": 23197
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.6139476961394765e-05,
+      "loss": 0.2388,
+      "step": 23198
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.612702366127023e-05,
+      "loss": 0.1744,
+      "step": 23199
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.61145703611457e-05,
+      "loss": 0.1174,
+      "step": 23200
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.610211706102116e-05,
+      "loss": 0.1307,
+      "step": 23201
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.6089663760896635e-05,
+      "loss": 0.1403,
+      "step": 23202
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.60772104607721e-05,
+      "loss": 0.0992,
+      "step": 23203
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.606475716064757e-05,
+      "loss": 0.1899,
+      "step": 23204
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.605230386052303e-05,
+      "loss": 0.1007,
+      "step": 23205
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.60398505603985e-05,
+      "loss": 0.1724,
+      "step": 23206
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.602739726027397e-05,
+      "loss": 0.1195,
+      "step": 23207
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.601494396014944e-05,
+      "loss": 0.1178,
+      "step": 23208
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.6002490660024904e-05,
+      "loss": 0.1705,
+      "step": 23209
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.599003735990037e-05,
+      "loss": 0.2015,
+      "step": 23210
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5977584059775836e-05,
+      "loss": 0.1645,
+      "step": 23211
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.596513075965131e-05,
+      "loss": 0.0925,
+      "step": 23212
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.595267745952677e-05,
+      "loss": 0.2354,
+      "step": 23213
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.594022415940224e-05,
+      "loss": 0.1049,
+      "step": 23214
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5927770859277706e-05,
+      "loss": 0.078,
+      "step": 23215
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.591531755915317e-05,
+      "loss": 0.2695,
+      "step": 23216
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.590286425902864e-05,
+      "loss": 0.1124,
+      "step": 23217
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5890410958904104e-05,
+      "loss": 0.1594,
+      "step": 23218
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.587795765877958e-05,
+      "loss": 0.141,
+      "step": 23219
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.586550435865504e-05,
+      "loss": 0.2299,
+      "step": 23220
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.585305105853051e-05,
+      "loss": 0.1163,
+      "step": 23221
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5840597758405974e-05,
+      "loss": 0.1219,
+      "step": 23222
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.582814445828144e-05,
+      "loss": 0.1226,
+      "step": 23223
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.581569115815691e-05,
+      "loss": 0.1453,
+      "step": 23224
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.580323785803237e-05,
+      "loss": 0.1034,
+      "step": 23225
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5790784557907845e-05,
+      "loss": 0.0855,
+      "step": 23226
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.577833125778331e-05,
+      "loss": 0.0572,
+      "step": 23227
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.576587795765878e-05,
+      "loss": 0.1783,
+      "step": 23228
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.575342465753424e-05,
+      "loss": 0.2602,
+      "step": 23229
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.574097135740971e-05,
+      "loss": 0.1568,
+      "step": 23230
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.572851805728518e-05,
+      "loss": 0.123,
+      "step": 23231
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.571606475716065e-05,
+      "loss": 0.0666,
+      "step": 23232
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.5703611457036113e-05,
+      "loss": 0.1317,
+      "step": 23233
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.569115815691158e-05,
+      "loss": 0.1255,
+      "step": 23234
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5678704856787045e-05,
+      "loss": 0.1349,
+      "step": 23235
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.566625155666251e-05,
+      "loss": 0.1628,
+      "step": 23236
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.565379825653798e-05,
+      "loss": 0.1464,
+      "step": 23237
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.564134495641344e-05,
+      "loss": 0.0812,
+      "step": 23238
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5628891656288916e-05,
+      "loss": 0.0969,
+      "step": 23239
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.561643835616438e-05,
+      "loss": 0.1171,
+      "step": 23240
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.560398505603985e-05,
+      "loss": 0.1943,
+      "step": 23241
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5591531755915314e-05,
+      "loss": 0.1438,
+      "step": 23242
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.557907845579078e-05,
+      "loss": 0.1007,
+      "step": 23243
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5566625155666246e-05,
+      "loss": 0.1832,
+      "step": 23244
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.555417185554171e-05,
+      "loss": 0.1164,
+      "step": 23245
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5541718555417184e-05,
+      "loss": 0.1315,
+      "step": 23246
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.552926525529265e-05,
+      "loss": 0.2041,
+      "step": 23247
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5516811955168116e-05,
+      "loss": 0.193,
+      "step": 23248
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.550435865504358e-05,
+      "loss": 0.0993,
+      "step": 23249
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.549190535491905e-05,
+      "loss": 0.105,
+      "step": 23250
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.547945205479452e-05,
+      "loss": 0.1256,
+      "step": 23251
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.546699875466998e-05,
+      "loss": 0.1322,
+      "step": 23252
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.545454545454545e-05,
+      "loss": 0.089,
+      "step": 23253
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.544209215442092e-05,
+      "loss": 0.2169,
+      "step": 23254
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5429638854296385e-05,
+      "loss": 0.2535,
+      "step": 23255
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.541718555417185e-05,
+      "loss": 0.1556,
+      "step": 23256
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5404732254047316e-05,
+      "loss": 0.1603,
+      "step": 23257
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.539227895392279e-05,
+      "loss": 0.2523,
+      "step": 23258
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.5379825653798255e-05,
+      "loss": 0.0753,
+      "step": 23259
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.536737235367372e-05,
+      "loss": 0.1468,
+      "step": 23260
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.535491905354919e-05,
+      "loss": 0.1781,
+      "step": 23261
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.534246575342465e-05,
+      "loss": 0.1796,
+      "step": 23262
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5330012453300126e-05,
+      "loss": 0.0817,
+      "step": 23263
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5317559153175585e-05,
+      "loss": 0.1103,
+      "step": 23264
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.530510585305106e-05,
+      "loss": 0.1042,
+      "step": 23265
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5292652552926524e-05,
+      "loss": 0.1844,
+      "step": 23266
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.528019925280199e-05,
+      "loss": 0.1575,
+      "step": 23267
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5267745952677455e-05,
+      "loss": 0.0877,
+      "step": 23268
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.525529265255292e-05,
+      "loss": 0.1127,
+      "step": 23269
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5242839352428394e-05,
+      "loss": 0.1541,
+      "step": 23270
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.523038605230386e-05,
+      "loss": 0.0623,
+      "step": 23271
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5217932752179326e-05,
+      "loss": 0.1848,
+      "step": 23272
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.520547945205479e-05,
+      "loss": 0.2099,
+      "step": 23273
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.519302615193026e-05,
+      "loss": 0.1383,
+      "step": 23274
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.518057285180573e-05,
+      "loss": 0.1007,
+      "step": 23275
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.516811955168119e-05,
+      "loss": 0.0883,
+      "step": 23276
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.515566625155666e-05,
+      "loss": 0.0661,
+      "step": 23277
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.514321295143213e-05,
+      "loss": 0.1187,
+      "step": 23278
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5130759651307594e-05,
+      "loss": 0.1906,
+      "step": 23279
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.511830635118306e-05,
+      "loss": 0.1546,
+      "step": 23280
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5105853051058526e-05,
+      "loss": 0.1674,
+      "step": 23281
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.509339975093399e-05,
+      "loss": 0.0965,
+      "step": 23282
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5080946450809465e-05,
+      "loss": 0.1841,
+      "step": 23283
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.5068493150684924e-05,
+      "loss": 0.0667,
+      "step": 23284
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.50560398505604e-05,
+      "loss": 0.1341,
+      "step": 23285
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.504358655043586e-05,
+      "loss": 0.1687,
+      "step": 23286
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.503113325031133e-05,
+      "loss": 0.1712,
+      "step": 23287
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.5018679950186795e-05,
+      "loss": 0.185,
+      "step": 23288
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.500622665006226e-05,
+      "loss": 0.1021,
+      "step": 23289
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.499377334993773e-05,
+      "loss": 0.0627,
+      "step": 23290
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.498132004981319e-05,
+      "loss": 0.1752,
+      "step": 23291
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.4968866749688665e-05,
+      "loss": 0.1834,
+      "step": 23292
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.495641344956413e-05,
+      "loss": 0.1601,
+      "step": 23293
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.49439601494396e-05,
+      "loss": 0.0724,
+      "step": 23294
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.493150684931506e-05,
+      "loss": 0.0871,
+      "step": 23295
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.491905354919053e-05,
+      "loss": 0.0705,
+      "step": 23296
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.4906600249066e-05,
+      "loss": 0.1083,
+      "step": 23297
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.489414694894147e-05,
+      "loss": 0.1982,
+      "step": 23298
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.4881693648816934e-05,
+      "loss": 0.1136,
+      "step": 23299
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.48692403486924e-05,
+      "loss": 0.0929,
+      "step": 23300
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.4856787048567865e-05,
+      "loss": 0.1427,
+      "step": 23301
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.484433374844334e-05,
+      "loss": 0.1201,
+      "step": 23302
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.48318804483188e-05,
+      "loss": 0.099,
+      "step": 23303
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.481942714819427e-05,
+      "loss": 0.2164,
+      "step": 23304
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.4806973848069736e-05,
+      "loss": 0.1562,
+      "step": 23305
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.47945205479452e-05,
+      "loss": 0.1201,
+      "step": 23306
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.478206724782067e-05,
+      "loss": 0.1142,
+      "step": 23307
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.4769613947696134e-05,
+      "loss": 0.1155,
+      "step": 23308
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.4757160647571607e-05,
+      "loss": 0.166,
+      "step": 23309
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.474470734744707e-05,
+      "loss": 0.2502,
+      "step": 23310
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.473225404732254e-05,
+      "loss": 0.1291,
+      "step": 23311
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.4719800747198004e-05,
+      "loss": 0.2376,
+      "step": 23312
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.470734744707347e-05,
+      "loss": 0.103,
+      "step": 23313
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.469489414694894e-05,
+      "loss": 0.1233,
+      "step": 23314
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.46824408468244e-05,
+      "loss": 0.2121,
+      "step": 23315
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.4669987546699875e-05,
+      "loss": 0.1744,
+      "step": 23316
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.465753424657534e-05,
+      "loss": 0.1708,
+      "step": 23317
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.464508094645081e-05,
+      "loss": 0.0912,
+      "step": 23318
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.463262764632627e-05,
+      "loss": 0.1451,
+      "step": 23319
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.462017434620174e-05,
+      "loss": 0.1701,
+      "step": 23320
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.4607721046077205e-05,
+      "loss": 0.1408,
+      "step": 23321
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.459526774595268e-05,
+      "loss": 0.1976,
+      "step": 23322
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.458281444582814e-05,
+      "loss": 0.1197,
+      "step": 23323
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.457036114570361e-05,
+      "loss": 0.0895,
+      "step": 23324
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.4557907845579075e-05,
+      "loss": 0.1846,
+      "step": 23325
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.454545454545454e-05,
+      "loss": 0.147,
+      "step": 23326
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.453300124533001e-05,
+      "loss": 0.0595,
+      "step": 23327
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.452054794520547e-05,
+      "loss": 0.1644,
+      "step": 23328
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.4508094645080946e-05,
+      "loss": 0.178,
+      "step": 23329
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.449564134495641e-05,
+      "loss": 0.1572,
+      "step": 23330
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.448318804483188e-05,
+      "loss": 0.1025,
+      "step": 23331
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.4470734744707344e-05,
+      "loss": 0.1031,
+      "step": 23332
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.445828144458281e-05,
+      "loss": 0.1278,
+      "step": 23333
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.444582814445828e-05,
+      "loss": 0.0606,
+      "step": 23334
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.443337484433374e-05,
+      "loss": 0.1414,
+      "step": 23335
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.4420921544209214e-05,
+      "loss": 0.1955,
+      "step": 23336
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 3.440846824408468e-05,
+      "loss": 0.1177,
+      "step": 23337
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.4396014943960146e-05,
+      "loss": 0.1273,
+      "step": 23338
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.438356164383561e-05,
+      "loss": 0.1048,
+      "step": 23339
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.437110834371108e-05,
+      "loss": 0.0875,
+      "step": 23340
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.435865504358655e-05,
+      "loss": 0.2168,
+      "step": 23341
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.434620174346201e-05,
+      "loss": 0.1261,
+      "step": 23342
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.433374844333748e-05,
+      "loss": 0.1182,
+      "step": 23343
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.432129514321295e-05,
+      "loss": 0.1479,
+      "step": 23344
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.4308841843088414e-05,
+      "loss": 0.0713,
+      "step": 23345
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.429638854296388e-05,
+      "loss": 0.0766,
+      "step": 23346
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.4283935242839346e-05,
+      "loss": 0.1318,
+      "step": 23347
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.427148194271482e-05,
+      "loss": 0.1701,
+      "step": 23348
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.4259028642590285e-05,
+      "loss": 0.1121,
+      "step": 23349
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.424657534246575e-05,
+      "loss": 0.1472,
+      "step": 23350
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.423412204234122e-05,
+      "loss": 0.124,
+      "step": 23351
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.422166874221668e-05,
+      "loss": 0.1088,
+      "step": 23352
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.4209215442092156e-05,
+      "loss": 0.2241,
+      "step": 23353
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.4196762141967615e-05,
+      "loss": 0.1582,
+      "step": 23354
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.418430884184309e-05,
+      "loss": 0.1515,
+      "step": 23355
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.4171855541718553e-05,
+      "loss": 0.1693,
+      "step": 23356
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.415940224159402e-05,
+      "loss": 0.0937,
+      "step": 23357
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.4146948941469485e-05,
+      "loss": 0.1081,
+      "step": 23358
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.413449564134495e-05,
+      "loss": 0.1154,
+      "step": 23359
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.4122042341220424e-05,
+      "loss": 0.1411,
+      "step": 23360
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.410958904109589e-05,
+      "loss": 0.1388,
+      "step": 23361
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.4097135740971356e-05,
+      "loss": 0.1188,
+      "step": 23362
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.408468244084682e-05,
+      "loss": 0.0873,
+      "step": 23363
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.407222914072229e-05,
+      "loss": 0.1065,
+      "step": 23364
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.4059775840597754e-05,
+      "loss": 0.0887,
+      "step": 23365
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.404732254047322e-05,
+      "loss": 0.2577,
+      "step": 23366
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.4034869240348686e-05,
+      "loss": 0.1037,
+      "step": 23367
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.402241594022416e-05,
+      "loss": 0.1802,
+      "step": 23368
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.4009962640099624e-05,
+      "loss": 0.1695,
+      "step": 23369
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.399750933997509e-05,
+      "loss": 0.0792,
+      "step": 23370
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.3985056039850556e-05,
+      "loss": 0.139,
+      "step": 23371
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.397260273972602e-05,
+      "loss": 0.1339,
+      "step": 23372
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.3960149439601495e-05,
+      "loss": 0.159,
+      "step": 23373
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.3947696139476954e-05,
+      "loss": 0.1442,
+      "step": 23374
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.393524283935243e-05,
+      "loss": 0.1409,
+      "step": 23375
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.392278953922789e-05,
+      "loss": 0.1817,
+      "step": 23376
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.391033623910336e-05,
+      "loss": 0.144,
+      "step": 23377
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.3897882938978825e-05,
+      "loss": 0.0716,
+      "step": 23378
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.388542963885429e-05,
+      "loss": 0.2069,
+      "step": 23379
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.387297633872976e-05,
+      "loss": 0.1125,
+      "step": 23380
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.386052303860523e-05,
+      "loss": 0.1203,
+      "step": 23381
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.3848069738480695e-05,
+      "loss": 0.1308,
+      "step": 23382
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.383561643835616e-05,
+      "loss": 0.143,
+      "step": 23383
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.382316313823163e-05,
+      "loss": 0.0913,
+      "step": 23384
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.38107098381071e-05,
+      "loss": 0.2307,
+      "step": 23385
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.379825653798256e-05,
+      "loss": 0.1339,
+      "step": 23386
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.378580323785803e-05,
+      "loss": 0.1955,
+      "step": 23387
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.37733499377335e-05,
+      "loss": 0.1063,
+      "step": 23388
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.3760896637608963e-05,
+      "loss": 0.1761,
+      "step": 23389
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.374844333748443e-05,
+      "loss": 0.1223,
+      "step": 23390
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.3735990037359895e-05,
+      "loss": 0.1686,
+      "step": 23391
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.372353673723537e-05,
+      "loss": 0.1056,
+      "step": 23392
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.3711083437110834e-05,
+      "loss": 0.1607,
+      "step": 23393
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.36986301369863e-05,
+      "loss": 0.1113,
+      "step": 23394
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.3686176836861766e-05,
+      "loss": 0.1787,
+      "step": 23395
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.367372353673723e-05,
+      "loss": 0.0904,
+      "step": 23396
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.36612702366127e-05,
+      "loss": 0.2059,
+      "step": 23397
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.3648816936488164e-05,
+      "loss": 0.1918,
+      "step": 23398
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.3636363636363636e-05,
+      "loss": 0.2095,
+      "step": 23399
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.36239103362391e-05,
+      "loss": 0.1135,
+      "step": 23400
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.361145703611457e-05,
+      "loss": 0.0736,
+      "step": 23401
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.3599003735990034e-05,
+      "loss": 0.1417,
+      "step": 23402
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.35865504358655e-05,
+      "loss": 0.0763,
+      "step": 23403
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.357409713574097e-05,
+      "loss": 0.225,
+      "step": 23404
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.356164383561643e-05,
+      "loss": 0.1025,
+      "step": 23405
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.3549190535491905e-05,
+      "loss": 0.1035,
+      "step": 23406
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.353673723536737e-05,
+      "loss": 0.1091,
+      "step": 23407
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.352428393524284e-05,
+      "loss": 0.0642,
+      "step": 23408
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.35118306351183e-05,
+      "loss": 0.0784,
+      "step": 23409
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.349937733499377e-05,
+      "loss": 0.406,
+      "step": 23410
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.3486924034869235e-05,
+      "loss": 0.2019,
+      "step": 23411
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.347447073474471e-05,
+      "loss": 0.1521,
+      "step": 23412
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.3462017434620167e-05,
+      "loss": 0.1296,
+      "step": 23413
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.344956413449564e-05,
+      "loss": 0.1271,
+      "step": 23414
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 3.3437110834371105e-05,
+      "loss": 0.1026,
+      "step": 23415
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.342465753424657e-05,
+      "loss": 0.1549,
+      "step": 23416
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.341220423412204e-05,
+      "loss": 0.1833,
+      "step": 23417
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.33997509339975e-05,
+      "loss": 0.1242,
+      "step": 23418
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.3387297633872976e-05,
+      "loss": 0.1013,
+      "step": 23419
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.337484433374844e-05,
+      "loss": 0.1193,
+      "step": 23420
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.336239103362391e-05,
+      "loss": 0.1559,
+      "step": 23421
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.3349937733499374e-05,
+      "loss": 0.2123,
+      "step": 23422
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.333748443337484e-05,
+      "loss": 0.0979,
+      "step": 23423
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.332503113325031e-05,
+      "loss": 0.1089,
+      "step": 23424
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.331257783312577e-05,
+      "loss": 0.0775,
+      "step": 23425
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.3300124533001244e-05,
+      "loss": 0.1511,
+      "step": 23426
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.328767123287671e-05,
+      "loss": 0.1016,
+      "step": 23427
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.3275217932752176e-05,
+      "loss": 0.0805,
+      "step": 23428
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.326276463262764e-05,
+      "loss": 0.1825,
+      "step": 23429
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.325031133250311e-05,
+      "loss": 0.1371,
+      "step": 23430
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.323785803237858e-05,
+      "loss": 0.153,
+      "step": 23431
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.3225404732254047e-05,
+      "loss": 0.1126,
+      "step": 23432
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.321295143212951e-05,
+      "loss": 0.1003,
+      "step": 23433
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.320049813200498e-05,
+      "loss": 0.0981,
+      "step": 23434
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.3188044831880444e-05,
+      "loss": 0.1479,
+      "step": 23435
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.317559153175592e-05,
+      "loss": 0.1345,
+      "step": 23436
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.3163138231631376e-05,
+      "loss": 0.128,
+      "step": 23437
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.315068493150685e-05,
+      "loss": 0.2567,
+      "step": 23438
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.3138231631382315e-05,
+      "loss": 0.0876,
+      "step": 23439
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.312577833125778e-05,
+      "loss": 0.098,
+      "step": 23440
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.311332503113325e-05,
+      "loss": 0.2323,
+      "step": 23441
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.310087173100871e-05,
+      "loss": 0.1887,
+      "step": 23442
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.3088418430884186e-05,
+      "loss": 0.133,
+      "step": 23443
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.307596513075965e-05,
+      "loss": 0.1037,
+      "step": 23444
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.306351183063512e-05,
+      "loss": 0.1483,
+      "step": 23445
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.305105853051058e-05,
+      "loss": 0.1214,
+      "step": 23446
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.303860523038605e-05,
+      "loss": 0.1442,
+      "step": 23447
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.3026151930261515e-05,
+      "loss": 0.2342,
+      "step": 23448
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.301369863013698e-05,
+      "loss": 0.1408,
+      "step": 23449
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.300124533001245e-05,
+      "loss": 0.1192,
+      "step": 23450
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.298879202988792e-05,
+      "loss": 0.0983,
+      "step": 23451
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.2976338729763386e-05,
+      "loss": 0.1014,
+      "step": 23452
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.296388542963885e-05,
+      "loss": 0.0638,
+      "step": 23453
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.295143212951432e-05,
+      "loss": 0.1682,
+      "step": 23454
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.2938978829389784e-05,
+      "loss": 0.1133,
+      "step": 23455
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.292652552926525e-05,
+      "loss": 0.1041,
+      "step": 23456
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.2914072229140716e-05,
+      "loss": 0.1103,
+      "step": 23457
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.290161892901619e-05,
+      "loss": 0.1408,
+      "step": 23458
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.2889165628891654e-05,
+      "loss": 0.1538,
+      "step": 23459
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.287671232876712e-05,
+      "loss": 0.1938,
+      "step": 23460
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.2864259028642586e-05,
+      "loss": 0.1512,
+      "step": 23461
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.285180572851805e-05,
+      "loss": 0.2353,
+      "step": 23462
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.2839352428393525e-05,
+      "loss": 0.1226,
+      "step": 23463
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.2826899128268984e-05,
+      "loss": 0.2432,
+      "step": 23464
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.281444582814446e-05,
+      "loss": 0.1017,
+      "step": 23465
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.280199252801992e-05,
+      "loss": 0.1676,
+      "step": 23466
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.278953922789539e-05,
+      "loss": 0.193,
+      "step": 23467
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.2777085927770854e-05,
+      "loss": 0.1205,
+      "step": 23468
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.276463262764632e-05,
+      "loss": 0.0957,
+      "step": 23469
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.275217932752179e-05,
+      "loss": 0.1812,
+      "step": 23470
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.273972602739726e-05,
+      "loss": 0.0678,
+      "step": 23471
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.2727272727272725e-05,
+      "loss": 0.1579,
+      "step": 23472
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.271481942714819e-05,
+      "loss": 0.1448,
+      "step": 23473
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.270236612702366e-05,
+      "loss": 0.1237,
+      "step": 23474
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.268991282689913e-05,
+      "loss": 0.069,
+      "step": 23475
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.267745952677459e-05,
+      "loss": 0.1019,
+      "step": 23476
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.266500622665006e-05,
+      "loss": 0.107,
+      "step": 23477
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.265255292652553e-05,
+      "loss": 0.1099,
+      "step": 23478
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.2640099626400993e-05,
+      "loss": 0.1344,
+      "step": 23479
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.262764632627646e-05,
+      "loss": 0.1244,
+      "step": 23480
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.2615193026151925e-05,
+      "loss": 0.142,
+      "step": 23481
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.26027397260274e-05,
+      "loss": 0.1221,
+      "step": 23482
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.2590286425902864e-05,
+      "loss": 0.1323,
+      "step": 23483
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.257783312577833e-05,
+      "loss": 0.1606,
+      "step": 23484
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.2565379825653796e-05,
+      "loss": 0.0847,
+      "step": 23485
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.255292652552926e-05,
+      "loss": 0.1007,
+      "step": 23486
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.2540473225404735e-05,
+      "loss": 0.1869,
+      "step": 23487
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.2528019925280194e-05,
+      "loss": 0.2821,
+      "step": 23488
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.2515566625155666e-05,
+      "loss": 0.21,
+      "step": 23489
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.250311332503113e-05,
+      "loss": 0.1043,
+      "step": 23490
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.24906600249066e-05,
+      "loss": 0.1312,
+      "step": 23491
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.2478206724782064e-05,
+      "loss": 0.0571,
+      "step": 23492
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.246575342465753e-05,
+      "loss": 0.1488,
+      "step": 23493
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.2453300124532996e-05,
+      "loss": 0.2059,
+      "step": 23494
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.244084682440847e-05,
+      "loss": 0.1117,
+      "step": 23495
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.242839352428393e-05,
+      "loss": 0.1759,
+      "step": 23496
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.24159402241594e-05,
+      "loss": 0.085,
+      "step": 23497
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.240348692403487e-05,
+      "loss": 0.1152,
+      "step": 23498
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.239103362391033e-05,
+      "loss": 0.1587,
+      "step": 23499
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.23785803237858e-05,
+      "loss": 0.1458,
+      "step": 23500
+    },
+    {
+      "epoch": 9.01,
+      "eval_cer": 0.10023216341182668,
+      "eval_loss": 0.39906588196754456,
+      "eval_runtime": 48.0028,
+      "eval_samples_per_second": 36.519,
+      "eval_steps_per_second": 4.583,
+      "eval_wer": 0.1573743042041906,
+      "step": 23500
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.2366127023661265e-05,
+      "loss": 0.138,
+      "step": 23501
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.235367372353674e-05,
+      "loss": 0.1414,
+      "step": 23502
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.2341220423412196e-05,
+      "loss": 0.0742,
+      "step": 23503
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.232876712328767e-05,
+      "loss": 0.1063,
+      "step": 23504
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.2316313823163135e-05,
+      "loss": 0.1269,
+      "step": 23505
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.23038605230386e-05,
+      "loss": 0.2042,
+      "step": 23506
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.229140722291407e-05,
+      "loss": 0.1743,
+      "step": 23507
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.227895392278953e-05,
+      "loss": 0.1904,
+      "step": 23508
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.2266500622665006e-05,
+      "loss": 0.1731,
+      "step": 23509
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.225404732254047e-05,
+      "loss": 0.0744,
+      "step": 23510
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.224159402241594e-05,
+      "loss": 0.0891,
+      "step": 23511
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.2229140722291403e-05,
+      "loss": 0.1126,
+      "step": 23512
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.221668742216687e-05,
+      "loss": 0.2202,
+      "step": 23513
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.220423412204234e-05,
+      "loss": 0.0996,
+      "step": 23514
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.21917808219178e-05,
+      "loss": 0.0727,
+      "step": 23515
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.2179327521793274e-05,
+      "loss": 0.0752,
+      "step": 23516
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.216687422166874e-05,
+      "loss": 0.129,
+      "step": 23517
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.2154420921544206e-05,
+      "loss": 0.089,
+      "step": 23518
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.214196762141967e-05,
+      "loss": 0.1546,
+      "step": 23519
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 3.212951432129514e-05,
+      "loss": 0.1096,
+      "step": 23520
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.211706102117061e-05,
+      "loss": 0.2634,
+      "step": 23521
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.2104607721046076e-05,
+      "loss": 0.1143,
+      "step": 23522
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.209215442092154e-05,
+      "loss": 0.0962,
+      "step": 23523
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.207970112079701e-05,
+      "loss": 0.1094,
+      "step": 23524
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.2067247820672474e-05,
+      "loss": 0.213,
+      "step": 23525
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.205479452054795e-05,
+      "loss": 0.1371,
+      "step": 23526
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.2042341220423406e-05,
+      "loss": 0.0948,
+      "step": 23527
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.202988792029888e-05,
+      "loss": 0.1152,
+      "step": 23528
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.2017434620174345e-05,
+      "loss": 0.0878,
+      "step": 23529
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.200498132004981e-05,
+      "loss": 0.0629,
+      "step": 23530
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.199252801992528e-05,
+      "loss": 0.1047,
+      "step": 23531
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.198007471980074e-05,
+      "loss": 0.1747,
+      "step": 23532
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.1967621419676215e-05,
+      "loss": 0.1679,
+      "step": 23533
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.195516811955168e-05,
+      "loss": 0.0725,
+      "step": 23534
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.194271481942715e-05,
+      "loss": 0.1182,
+      "step": 23535
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.193026151930261e-05,
+      "loss": 0.1269,
+      "step": 23536
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.191780821917808e-05,
+      "loss": 0.0795,
+      "step": 23537
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.1905354919053545e-05,
+      "loss": 0.187,
+      "step": 23538
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.189290161892901e-05,
+      "loss": 0.1143,
+      "step": 23539
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.188044831880448e-05,
+      "loss": 0.0906,
+      "step": 23540
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.186799501867995e-05,
+      "loss": 0.1414,
+      "step": 23541
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.1855541718555416e-05,
+      "loss": 0.0688,
+      "step": 23542
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.184308841843088e-05,
+      "loss": 0.1402,
+      "step": 23543
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.183063511830635e-05,
+      "loss": 0.17,
+      "step": 23544
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.1818181818181814e-05,
+      "loss": 0.1113,
+      "step": 23545
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.1805728518057286e-05,
+      "loss": 0.1101,
+      "step": 23546
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.1793275217932745e-05,
+      "loss": 0.0677,
+      "step": 23547
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.178082191780822e-05,
+      "loss": 0.0648,
+      "step": 23548
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.1768368617683684e-05,
+      "loss": 0.1151,
+      "step": 23549
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.175591531755915e-05,
+      "loss": 0.108,
+      "step": 23550
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.1743462017434616e-05,
+      "loss": 0.192,
+      "step": 23551
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.173100871731008e-05,
+      "loss": 0.0914,
+      "step": 23552
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.1718555417185555e-05,
+      "loss": 0.1063,
+      "step": 23553
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.1706102117061014e-05,
+      "loss": 0.1359,
+      "step": 23554
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.1693648816936487e-05,
+      "loss": 0.0604,
+      "step": 23555
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.168119551681195e-05,
+      "loss": 0.2264,
+      "step": 23556
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.166874221668742e-05,
+      "loss": 0.1595,
+      "step": 23557
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.1656288916562884e-05,
+      "loss": 0.0779,
+      "step": 23558
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.164383561643835e-05,
+      "loss": 0.1235,
+      "step": 23559
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.163138231631382e-05,
+      "loss": 0.0815,
+      "step": 23560
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.161892901618929e-05,
+      "loss": 0.0995,
+      "step": 23561
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.1606475716064755e-05,
+      "loss": 0.0709,
+      "step": 23562
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.159402241594022e-05,
+      "loss": 0.2818,
+      "step": 23563
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.158156911581569e-05,
+      "loss": 0.2135,
+      "step": 23564
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.156911581569116e-05,
+      "loss": 0.1006,
+      "step": 23565
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.155666251556662e-05,
+      "loss": 0.1115,
+      "step": 23566
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.154420921544209e-05,
+      "loss": 0.0595,
+      "step": 23567
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.153175591531756e-05,
+      "loss": 0.16,
+      "step": 23568
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.151930261519302e-05,
+      "loss": 0.2101,
+      "step": 23569
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.150684931506849e-05,
+      "loss": 0.1254,
+      "step": 23570
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.1494396014943955e-05,
+      "loss": 0.1171,
+      "step": 23571
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.148194271481943e-05,
+      "loss": 0.1365,
+      "step": 23572
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.1469489414694894e-05,
+      "loss": 0.1078,
+      "step": 23573
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.145703611457036e-05,
+      "loss": 0.1192,
+      "step": 23574
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.1444582814445826e-05,
+      "loss": 0.2385,
+      "step": 23575
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.143212951432129e-05,
+      "loss": 0.1615,
+      "step": 23576
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.141967621419676e-05,
+      "loss": 0.1072,
+      "step": 23577
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.1407222914072224e-05,
+      "loss": 0.119,
+      "step": 23578
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.139476961394769e-05,
+      "loss": 0.0735,
+      "step": 23579
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.138231631382316e-05,
+      "loss": 0.0917,
+      "step": 23580
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.136986301369863e-05,
+      "loss": 0.1801,
+      "step": 23581
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.1357409713574094e-05,
+      "loss": 0.1748,
+      "step": 23582
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.134495641344956e-05,
+      "loss": 0.1135,
+      "step": 23583
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.1332503113325026e-05,
+      "loss": 0.159,
+      "step": 23584
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.13200498132005e-05,
+      "loss": 0.1411,
+      "step": 23585
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.130759651307596e-05,
+      "loss": 0.0557,
+      "step": 23586
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.129514321295143e-05,
+      "loss": 0.0891,
+      "step": 23587
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.12826899128269e-05,
+      "loss": 0.1867,
+      "step": 23588
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.127023661270236e-05,
+      "loss": 0.1721,
+      "step": 23589
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.125778331257783e-05,
+      "loss": 0.153,
+      "step": 23590
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.1245330012453294e-05,
+      "loss": 0.1905,
+      "step": 23591
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.123287671232877e-05,
+      "loss": 0.0774,
+      "step": 23592
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.122042341220423e-05,
+      "loss": 0.0994,
+      "step": 23593
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.12079701120797e-05,
+      "loss": 0.1175,
+      "step": 23594
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.1195516811955165e-05,
+      "loss": 0.1025,
+      "step": 23595
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.118306351183063e-05,
+      "loss": 0.2502,
+      "step": 23596
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.1170610211706104e-05,
+      "loss": 0.1887,
+      "step": 23597
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.115815691158156e-05,
+      "loss": 0.1332,
+      "step": 23598
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.1145703611457036e-05,
+      "loss": 0.1025,
+      "step": 23599
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.11332503113325e-05,
+      "loss": 0.1981,
+      "step": 23600
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.112079701120797e-05,
+      "loss": 0.1728,
+      "step": 23601
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.1108343711083433e-05,
+      "loss": 0.1557,
+      "step": 23602
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.10958904109589e-05,
+      "loss": 0.1021,
+      "step": 23603
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.108343711083437e-05,
+      "loss": 0.0781,
+      "step": 23604
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.107098381070984e-05,
+      "loss": 0.1206,
+      "step": 23605
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.1058530510585304e-05,
+      "loss": 0.0833,
+      "step": 23606
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.104607721046077e-05,
+      "loss": 0.1714,
+      "step": 23607
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.1033623910336236e-05,
+      "loss": 0.139,
+      "step": 23608
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.10211706102117e-05,
+      "loss": 0.0948,
+      "step": 23609
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.100871731008717e-05,
+      "loss": 0.0866,
+      "step": 23610
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.099626400996264e-05,
+      "loss": 0.0862,
+      "step": 23611
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.0983810709838106e-05,
+      "loss": 0.093,
+      "step": 23612
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.097135740971357e-05,
+      "loss": 0.1364,
+      "step": 23613
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.095890410958904e-05,
+      "loss": 0.0974,
+      "step": 23614
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.0946450809464504e-05,
+      "loss": 0.1755,
+      "step": 23615
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.093399750933998e-05,
+      "loss": 0.09,
+      "step": 23616
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.0921544209215436e-05,
+      "loss": 0.0538,
+      "step": 23617
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.090909090909091e-05,
+      "loss": 0.0941,
+      "step": 23618
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.0896637608966375e-05,
+      "loss": 0.1348,
+      "step": 23619
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.088418430884184e-05,
+      "loss": 0.1199,
+      "step": 23620
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.087173100871731e-05,
+      "loss": 0.1831,
+      "step": 23621
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.085927770859277e-05,
+      "loss": 0.0879,
+      "step": 23622
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.084682440846824e-05,
+      "loss": 0.0836,
+      "step": 23623
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 3.083437110834371e-05,
+      "loss": 0.0988,
+      "step": 23624
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.082191780821917e-05,
+      "loss": 0.1379,
+      "step": 23625
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.080946450809464e-05,
+      "loss": 0.1236,
+      "step": 23626
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.079701120797011e-05,
+      "loss": 0.0914,
+      "step": 23627
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.0784557907845575e-05,
+      "loss": 0.0882,
+      "step": 23628
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.077210460772104e-05,
+      "loss": 0.0669,
+      "step": 23629
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.075965130759651e-05,
+      "loss": 0.0765,
+      "step": 23630
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.074719800747198e-05,
+      "loss": 0.1422,
+      "step": 23631
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.0734744707347446e-05,
+      "loss": 0.1169,
+      "step": 23632
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.072229140722291e-05,
+      "loss": 0.1528,
+      "step": 23633
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.070983810709838e-05,
+      "loss": 0.0841,
+      "step": 23634
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.0697384806973843e-05,
+      "loss": 0.1236,
+      "step": 23635
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.0684931506849316e-05,
+      "loss": 0.1015,
+      "step": 23636
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.0672478206724775e-05,
+      "loss": 0.0879,
+      "step": 23637
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.066002490660025e-05,
+      "loss": 0.234,
+      "step": 23638
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.0647571606475714e-05,
+      "loss": 0.1309,
+      "step": 23639
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.063511830635118e-05,
+      "loss": 0.1367,
+      "step": 23640
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.0622665006226646e-05,
+      "loss": 0.1351,
+      "step": 23641
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.061021170610211e-05,
+      "loss": 0.0601,
+      "step": 23642
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.0597758405977585e-05,
+      "loss": 0.0955,
+      "step": 23643
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.058530510585305e-05,
+      "loss": 0.1146,
+      "step": 23644
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.0572851805728516e-05,
+      "loss": 0.1821,
+      "step": 23645
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.056039850560398e-05,
+      "loss": 0.0897,
+      "step": 23646
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.054794520547945e-05,
+      "loss": 0.0901,
+      "step": 23647
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.053549190535492e-05,
+      "loss": 0.1086,
+      "step": 23648
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.052303860523038e-05,
+      "loss": 0.1112,
+      "step": 23649
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.051058530510585e-05,
+      "loss": 0.2664,
+      "step": 23650
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.049813200498132e-05,
+      "loss": 0.1098,
+      "step": 23651
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0485678704856785e-05,
+      "loss": 0.1655,
+      "step": 23652
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.047322540473225e-05,
+      "loss": 0.1225,
+      "step": 23653
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0460772104607717e-05,
+      "loss": 0.0739,
+      "step": 23654
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0448318804483186e-05,
+      "loss": 0.1292,
+      "step": 23655
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0435865504358655e-05,
+      "loss": 0.1262,
+      "step": 23656
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0423412204234118e-05,
+      "loss": 0.1323,
+      "step": 23657
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0410958904109587e-05,
+      "loss": 0.1158,
+      "step": 23658
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0398505603985053e-05,
+      "loss": 0.0943,
+      "step": 23659
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.038605230386052e-05,
+      "loss": 0.129,
+      "step": 23660
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0373599003735985e-05,
+      "loss": 0.0878,
+      "step": 23661
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0361145703611454e-05,
+      "loss": 0.0942,
+      "step": 23662
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0348692403486924e-05,
+      "loss": 0.1602,
+      "step": 23663
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0336239103362386e-05,
+      "loss": 0.0867,
+      "step": 23664
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0323785803237856e-05,
+      "loss": 0.1134,
+      "step": 23665
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.031133250311332e-05,
+      "loss": 0.1386,
+      "step": 23666
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.029887920298879e-05,
+      "loss": 0.0934,
+      "step": 23667
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0286425902864254e-05,
+      "loss": 0.1018,
+      "step": 23668
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0273972602739723e-05,
+      "loss": 0.1389,
+      "step": 23669
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0261519302615192e-05,
+      "loss": 0.0975,
+      "step": 23670
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0249066002490658e-05,
+      "loss": 0.0921,
+      "step": 23671
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0236612702366124e-05,
+      "loss": 0.0903,
+      "step": 23672
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.022415940224159e-05,
+      "loss": 0.0531,
+      "step": 23673
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.021170610211706e-05,
+      "loss": 0.1885,
+      "step": 23674
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0199252801992525e-05,
+      "loss": 0.222,
+      "step": 23675
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.018679950186799e-05,
+      "loss": 0.0912,
+      "step": 23676
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.0174346201743457e-05,
+      "loss": 0.1071,
+      "step": 23677
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.0161892901618927e-05,
+      "loss": 0.1121,
+      "step": 23678
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.0149439601494396e-05,
+      "loss": 0.1045,
+      "step": 23679
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.013698630136986e-05,
+      "loss": 0.1027,
+      "step": 23680
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.0124533001245328e-05,
+      "loss": 0.076,
+      "step": 23681
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.0112079701120794e-05,
+      "loss": 0.1415,
+      "step": 23682
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.0099626400996263e-05,
+      "loss": 0.1818,
+      "step": 23683
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.0087173100871726e-05,
+      "loss": 0.0891,
+      "step": 23684
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.0074719800747195e-05,
+      "loss": 0.1163,
+      "step": 23685
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.0062266500622664e-05,
+      "loss": 0.0898,
+      "step": 23686
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.004981320049813e-05,
+      "loss": 0.087,
+      "step": 23687
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.0037359900373596e-05,
+      "loss": 0.1564,
+      "step": 23688
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.0024906600249062e-05,
+      "loss": 0.2425,
+      "step": 23689
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 3.001245330012453e-05,
+      "loss": 0.1612,
+      "step": 23690
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 0.1278,
+      "step": 23691
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.9987546699875463e-05,
+      "loss": 0.094,
+      "step": 23692
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.9975093399750933e-05,
+      "loss": 0.1017,
+      "step": 23693
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.99626400996264e-05,
+      "loss": 0.2275,
+      "step": 23694
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.9950186799501868e-05,
+      "loss": 0.0874,
+      "step": 23695
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.993773349937733e-05,
+      "loss": 0.1055,
+      "step": 23696
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.99252801992528e-05,
+      "loss": 0.0629,
+      "step": 23697
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.9912826899128266e-05,
+      "loss": 0.1,
+      "step": 23698
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.9900373599003735e-05,
+      "loss": 0.1269,
+      "step": 23699
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.9887920298879198e-05,
+      "loss": 0.1599,
+      "step": 23700
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.9875466998754667e-05,
+      "loss": 0.1266,
+      "step": 23701
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 2.9863013698630136e-05,
+      "loss": 0.1095,
+      "step": 23702
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9850560398505602e-05,
+      "loss": 0.0949,
+      "step": 23703
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9838107098381068e-05,
+      "loss": 0.2067,
+      "step": 23704
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9825653798256534e-05,
+      "loss": 0.1458,
+      "step": 23705
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9813200498132003e-05,
+      "loss": 0.1534,
+      "step": 23706
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9800747198007473e-05,
+      "loss": 0.1458,
+      "step": 23707
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9788293897882935e-05,
+      "loss": 0.1502,
+      "step": 23708
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9775840597758405e-05,
+      "loss": 0.0965,
+      "step": 23709
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.976338729763387e-05,
+      "loss": 0.0702,
+      "step": 23710
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.975093399750934e-05,
+      "loss": 0.1043,
+      "step": 23711
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9738480697384803e-05,
+      "loss": 0.0725,
+      "step": 23712
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9726027397260272e-05,
+      "loss": 0.0767,
+      "step": 23713
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9713574097135738e-05,
+      "loss": 0.2524,
+      "step": 23714
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9701120797011204e-05,
+      "loss": 0.1751,
+      "step": 23715
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.968866749688667e-05,
+      "loss": 0.0773,
+      "step": 23716
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.967621419676214e-05,
+      "loss": 0.1364,
+      "step": 23717
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.966376089663761e-05,
+      "loss": 0.1835,
+      "step": 23718
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.965130759651307e-05,
+      "loss": 0.1766,
+      "step": 23719
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.963885429638854e-05,
+      "loss": 0.1023,
+      "step": 23720
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9626400996264006e-05,
+      "loss": 0.1493,
+      "step": 23721
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9613947696139476e-05,
+      "loss": 0.0826,
+      "step": 23722
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9601494396014938e-05,
+      "loss": 0.0698,
+      "step": 23723
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9589041095890407e-05,
+      "loss": 0.0576,
+      "step": 23724
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9576587795765877e-05,
+      "loss": 0.1505,
+      "step": 23725
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9564134495641343e-05,
+      "loss": 0.1531,
+      "step": 23726
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.955168119551681e-05,
+      "loss": 0.1823,
+      "step": 23727
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 2.9539227895392275e-05,
+      "loss": 0.1058,
+      "step": 23728
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9526774595267744e-05,
+      "loss": 0.0744,
+      "step": 23729
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9514321295143213e-05,
+      "loss": 0.0718,
+      "step": 23730
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9501867995018676e-05,
+      "loss": 0.0755,
+      "step": 23731
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9489414694894145e-05,
+      "loss": 0.1453,
+      "step": 23732
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.947696139476961e-05,
+      "loss": 0.1931,
+      "step": 23733
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.946450809464508e-05,
+      "loss": 0.1622,
+      "step": 23734
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9452054794520543e-05,
+      "loss": 0.0883,
+      "step": 23735
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9439601494396012e-05,
+      "loss": 0.0603,
+      "step": 23736
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9427148194271478e-05,
+      "loss": 0.1332,
+      "step": 23737
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9414694894146948e-05,
+      "loss": 0.176,
+      "step": 23738
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.940224159402241e-05,
+      "loss": 0.0896,
+      "step": 23739
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.938978829389788e-05,
+      "loss": 0.1523,
+      "step": 23740
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.937733499377335e-05,
+      "loss": 0.1538,
+      "step": 23741
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9364881693648815e-05,
+      "loss": 0.0955,
+      "step": 23742
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.935242839352428e-05,
+      "loss": 0.0797,
+      "step": 23743
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9339975093399747e-05,
+      "loss": 0.14,
+      "step": 23744
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9327521793275216e-05,
+      "loss": 0.1195,
+      "step": 23745
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9315068493150685e-05,
+      "loss": 0.1464,
+      "step": 23746
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9302615193026148e-05,
+      "loss": 0.0777,
+      "step": 23747
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9290161892901617e-05,
+      "loss": 0.0589,
+      "step": 23748
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9277708592777083e-05,
+      "loss": 0.1349,
+      "step": 23749
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9265255292652553e-05,
+      "loss": 0.17,
+      "step": 23750
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9252801992528015e-05,
+      "loss": 0.1053,
+      "step": 23751
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9240348692403484e-05,
+      "loss": 0.1865,
+      "step": 23752
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.9227895392278954e-05,
+      "loss": 0.0865,
+      "step": 23753
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 2.921544209215442e-05,
+      "loss": 0.0662,
+      "step": 23754
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9202988792029886e-05,
+      "loss": 0.0875,
+      "step": 23755
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.919053549190535e-05,
+      "loss": 0.0601,
+      "step": 23756
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.917808219178082e-05,
+      "loss": 0.1688,
+      "step": 23757
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9165628891656287e-05,
+      "loss": 0.1251,
+      "step": 23758
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9153175591531753e-05,
+      "loss": 0.0891,
+      "step": 23759
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.914072229140722e-05,
+      "loss": 0.0773,
+      "step": 23760
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9128268991282688e-05,
+      "loss": 0.0927,
+      "step": 23761
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9115815691158157e-05,
+      "loss": 0.1705,
+      "step": 23762
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.910336239103362e-05,
+      "loss": 0.1539,
+      "step": 23763
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.909090909090909e-05,
+      "loss": 0.095,
+      "step": 23764
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9078455790784555e-05,
+      "loss": 0.2156,
+      "step": 23765
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.906600249066002e-05,
+      "loss": 0.0813,
+      "step": 23766
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9053549190535487e-05,
+      "loss": 0.0758,
+      "step": 23767
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9041095890410956e-05,
+      "loss": 0.0591,
+      "step": 23768
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9028642590286426e-05,
+      "loss": 0.1241,
+      "step": 23769
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.901618929016189e-05,
+      "loss": 0.2337,
+      "step": 23770
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.9003735990037358e-05,
+      "loss": 0.0904,
+      "step": 23771
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.8991282689912824e-05,
+      "loss": 0.0738,
+      "step": 23772
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.8978829389788293e-05,
+      "loss": 0.0864,
+      "step": 23773
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.8966376089663756e-05,
+      "loss": 0.0807,
+      "step": 23774
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.8953922789539225e-05,
+      "loss": 0.2139,
+      "step": 23775
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.8941469489414694e-05,
+      "loss": 0.1798,
+      "step": 23776
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.892901618929016e-05,
+      "loss": 0.1343,
+      "step": 23777
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.8916562889165626e-05,
+      "loss": 0.1451,
+      "step": 23778
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.8904109589041092e-05,
+      "loss": 0.0826,
+      "step": 23779
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.889165628891656e-05,
+      "loss": 0.1237,
+      "step": 23780
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 2.8879202988792027e-05,
+      "loss": 0.1775,
+      "step": 23781
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8866749688667493e-05,
+      "loss": 0.1267,
+      "step": 23782
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.885429638854296e-05,
+      "loss": 0.2079,
+      "step": 23783
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.884184308841843e-05,
+      "loss": 0.0729,
+      "step": 23784
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8829389788293898e-05,
+      "loss": 0.125,
+      "step": 23785
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.881693648816936e-05,
+      "loss": 0.1106,
+      "step": 23786
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.880448318804483e-05,
+      "loss": 0.0924,
+      "step": 23787
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8792029887920296e-05,
+      "loss": 0.2578,
+      "step": 23788
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8779576587795765e-05,
+      "loss": 0.1626,
+      "step": 23789
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8767123287671228e-05,
+      "loss": 0.1012,
+      "step": 23790
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8754669987546697e-05,
+      "loss": 0.1708,
+      "step": 23791
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8742216687422166e-05,
+      "loss": 0.0819,
+      "step": 23792
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8729763387297632e-05,
+      "loss": 0.0701,
+      "step": 23793
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8717310087173098e-05,
+      "loss": 0.2051,
+      "step": 23794
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8704856787048564e-05,
+      "loss": 0.132,
+      "step": 23795
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8692403486924033e-05,
+      "loss": 0.1083,
+      "step": 23796
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.86799501867995e-05,
+      "loss": 0.1113,
+      "step": 23797
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8667496886674965e-05,
+      "loss": 0.1124,
+      "step": 23798
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8655043586550435e-05,
+      "loss": 0.0617,
+      "step": 23799
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.86425902864259e-05,
+      "loss": 0.254,
+      "step": 23800
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.863013698630137e-05,
+      "loss": 0.1309,
+      "step": 23801
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8617683686176832e-05,
+      "loss": 0.1594,
+      "step": 23802
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8605230386052302e-05,
+      "loss": 0.0814,
+      "step": 23803
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8592777085927768e-05,
+      "loss": 0.104,
+      "step": 23804
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.8580323785803237e-05,
+      "loss": 0.0825,
+      "step": 23805
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.85678704856787e-05,
+      "loss": 0.1105,
+      "step": 23806
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 2.855541718555417e-05,
+      "loss": 0.1035,
+      "step": 23807
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8542963885429638e-05,
+      "loss": 0.1328,
+      "step": 23808
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8530510585305104e-05,
+      "loss": 0.121,
+      "step": 23809
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.851805728518057e-05,
+      "loss": 0.0937,
+      "step": 23810
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8505603985056036e-05,
+      "loss": 0.137,
+      "step": 23811
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8493150684931505e-05,
+      "loss": 0.1972,
+      "step": 23812
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8480697384806975e-05,
+      "loss": 0.1537,
+      "step": 23813
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8468244084682437e-05,
+      "loss": 0.1536,
+      "step": 23814
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8455790784557907e-05,
+      "loss": 0.1548,
+      "step": 23815
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8443337484433373e-05,
+      "loss": 0.054,
+      "step": 23816
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8430884184308842e-05,
+      "loss": 0.1156,
+      "step": 23817
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8418430884184305e-05,
+      "loss": 0.0844,
+      "step": 23818
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8405977584059774e-05,
+      "loss": 0.2203,
+      "step": 23819
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.839352428393524e-05,
+      "loss": 0.0866,
+      "step": 23820
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8381070983810706e-05,
+      "loss": 0.0682,
+      "step": 23821
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.836861768368617e-05,
+      "loss": 0.1015,
+      "step": 23822
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.835616438356164e-05,
+      "loss": 0.0753,
+      "step": 23823
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.834371108343711e-05,
+      "loss": 0.0875,
+      "step": 23824
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8331257783312573e-05,
+      "loss": 0.1319,
+      "step": 23825
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8318804483188042e-05,
+      "loss": 0.1886,
+      "step": 23826
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8306351183063508e-05,
+      "loss": 0.1547,
+      "step": 23827
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8293897882938978e-05,
+      "loss": 0.0976,
+      "step": 23828
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.828144458281444e-05,
+      "loss": 0.072,
+      "step": 23829
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.826899128268991e-05,
+      "loss": 0.0978,
+      "step": 23830
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.825653798256538e-05,
+      "loss": 0.1292,
+      "step": 23831
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.8244084682440845e-05,
+      "loss": 0.1096,
+      "step": 23832
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 2.823163138231631e-05,
+      "loss": 0.094,
+      "step": 23833
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8219178082191777e-05,
+      "loss": 0.1881,
+      "step": 23834
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8206724782067246e-05,
+      "loss": 0.0634,
+      "step": 23835
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8194271481942715e-05,
+      "loss": 0.067,
+      "step": 23836
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8181818181818178e-05,
+      "loss": 0.1163,
+      "step": 23837
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8169364881693647e-05,
+      "loss": 0.223,
+      "step": 23838
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8156911581569113e-05,
+      "loss": 0.1374,
+      "step": 23839
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8144458281444582e-05,
+      "loss": 0.1417,
+      "step": 23840
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8132004981320045e-05,
+      "loss": 0.1163,
+      "step": 23841
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8119551681195514e-05,
+      "loss": 0.0744,
+      "step": 23842
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.810709838107098e-05,
+      "loss": 0.1153,
+      "step": 23843
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.809464508094645e-05,
+      "loss": 0.1136,
+      "step": 23844
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8082191780821912e-05,
+      "loss": 0.1144,
+      "step": 23845
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.806973848069738e-05,
+      "loss": 0.0847,
+      "step": 23846
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.805728518057285e-05,
+      "loss": 0.1086,
+      "step": 23847
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8044831880448317e-05,
+      "loss": 0.1253,
+      "step": 23848
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8032378580323783e-05,
+      "loss": 0.1698,
+      "step": 23849
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.801992528019925e-05,
+      "loss": 0.1318,
+      "step": 23850
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.8007471980074718e-05,
+      "loss": 0.1983,
+      "step": 23851
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.7995018679950187e-05,
+      "loss": 0.1018,
+      "step": 23852
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.798256537982565e-05,
+      "loss": 0.1324,
+      "step": 23853
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.797011207970112e-05,
+      "loss": 0.0774,
+      "step": 23854
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.7957658779576585e-05,
+      "loss": 0.0813,
+      "step": 23855
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.7945205479452054e-05,
+      "loss": 0.1036,
+      "step": 23856
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.7932752179327517e-05,
+      "loss": 0.1333,
+      "step": 23857
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.7920298879202986e-05,
+      "loss": 0.1139,
+      "step": 23858
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 2.7907845579078456e-05,
+      "loss": 0.07,
+      "step": 23859
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.789539227895392e-05,
+      "loss": 0.1251,
+      "step": 23860
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7882938978829388e-05,
+      "loss": 0.0973,
+      "step": 23861
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7870485678704854e-05,
+      "loss": 0.1059,
+      "step": 23862
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7858032378580323e-05,
+      "loss": 0.2753,
+      "step": 23863
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.784557907845579e-05,
+      "loss": 0.1331,
+      "step": 23864
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7833125778331255e-05,
+      "loss": 0.136,
+      "step": 23865
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.782067247820672e-05,
+      "loss": 0.158,
+      "step": 23866
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.780821917808219e-05,
+      "loss": 0.0792,
+      "step": 23867
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.779576587795766e-05,
+      "loss": 0.098,
+      "step": 23868
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7783312577833122e-05,
+      "loss": 0.2865,
+      "step": 23869
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.777085927770859e-05,
+      "loss": 0.1943,
+      "step": 23870
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7758405977584057e-05,
+      "loss": 0.1189,
+      "step": 23871
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7745952677459523e-05,
+      "loss": 0.0752,
+      "step": 23872
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.773349937733499e-05,
+      "loss": 0.1288,
+      "step": 23873
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.772104607721046e-05,
+      "loss": 0.091,
+      "step": 23874
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7708592777085928e-05,
+      "loss": 0.1324,
+      "step": 23875
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.769613947696139e-05,
+      "loss": 0.1246,
+      "step": 23876
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.768368617683686e-05,
+      "loss": 0.0796,
+      "step": 23877
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7671232876712326e-05,
+      "loss": 0.1486,
+      "step": 23878
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7658779576587795e-05,
+      "loss": 0.0883,
+      "step": 23879
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7646326276463257e-05,
+      "loss": 0.0763,
+      "step": 23880
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7633872976338727e-05,
+      "loss": 0.1074,
+      "step": 23881
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7621419676214196e-05,
+      "loss": 0.1919,
+      "step": 23882
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7608966376089662e-05,
+      "loss": 0.0866,
+      "step": 23883
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7596513075965128e-05,
+      "loss": 0.0907,
+      "step": 23884
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 2.7584059775840594e-05,
+      "loss": 0.1143,
+      "step": 23885
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7571606475716063e-05,
+      "loss": 0.097,
+      "step": 23886
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.755915317559153e-05,
+      "loss": 0.0641,
+      "step": 23887
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7546699875466995e-05,
+      "loss": 0.1886,
+      "step": 23888
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.753424657534246e-05,
+      "loss": 0.1108,
+      "step": 23889
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.752179327521793e-05,
+      "loss": 0.1176,
+      "step": 23890
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.75093399750934e-05,
+      "loss": 0.0981,
+      "step": 23891
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7496886674968862e-05,
+      "loss": 0.1436,
+      "step": 23892
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7484433374844332e-05,
+      "loss": 0.07,
+      "step": 23893
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7471980074719798e-05,
+      "loss": 0.1288,
+      "step": 23894
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7459526774595267e-05,
+      "loss": 0.173,
+      "step": 23895
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.744707347447073e-05,
+      "loss": 0.1148,
+      "step": 23896
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.74346201743462e-05,
+      "loss": 0.0984,
+      "step": 23897
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7422166874221668e-05,
+      "loss": 0.0733,
+      "step": 23898
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7409713574097134e-05,
+      "loss": 0.077,
+      "step": 23899
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.73972602739726e-05,
+      "loss": 0.1292,
+      "step": 23900
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7384806973848066e-05,
+      "loss": 0.1693,
+      "step": 23901
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7372353673723535e-05,
+      "loss": 0.0911,
+      "step": 23902
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7359900373599e-05,
+      "loss": 0.2479,
+      "step": 23903
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7347447073474467e-05,
+      "loss": 0.0667,
+      "step": 23904
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7334993773349937e-05,
+      "loss": 0.1788,
+      "step": 23905
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7322540473225403e-05,
+      "loss": 0.0883,
+      "step": 23906
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7310087173100872e-05,
+      "loss": 0.1674,
+      "step": 23907
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7297633872976334e-05,
+      "loss": 0.0769,
+      "step": 23908
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.7285180572851804e-05,
+      "loss": 0.0706,
+      "step": 23909
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.727272727272727e-05,
+      "loss": 0.0538,
+      "step": 23910
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 2.726027397260274e-05,
+      "loss": 0.065,
+      "step": 23911
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.72478206724782e-05,
+      "loss": 0.109,
+      "step": 23912
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.723536737235367e-05,
+      "loss": 0.183,
+      "step": 23913
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.722291407222914e-05,
+      "loss": 0.1599,
+      "step": 23914
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7210460772104606e-05,
+      "loss": 0.0814,
+      "step": 23915
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7198007471980072e-05,
+      "loss": 0.1199,
+      "step": 23916
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7185554171855538e-05,
+      "loss": 0.0776,
+      "step": 23917
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7173100871731007e-05,
+      "loss": 0.0564,
+      "step": 23918
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7160647571606477e-05,
+      "loss": 0.2019,
+      "step": 23919
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.714819427148194e-05,
+      "loss": 0.1258,
+      "step": 23920
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.713574097135741e-05,
+      "loss": 0.112,
+      "step": 23921
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7123287671232875e-05,
+      "loss": 0.1384,
+      "step": 23922
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7110834371108344e-05,
+      "loss": 0.0842,
+      "step": 23923
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7098381070983807e-05,
+      "loss": 0.0905,
+      "step": 23924
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7085927770859276e-05,
+      "loss": 0.1625,
+      "step": 23925
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7073474470734742e-05,
+      "loss": 0.1964,
+      "step": 23926
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7061021170610208e-05,
+      "loss": 0.1012,
+      "step": 23927
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7048567870485674e-05,
+      "loss": 0.1212,
+      "step": 23928
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7036114570361143e-05,
+      "loss": 0.0768,
+      "step": 23929
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7023661270236612e-05,
+      "loss": 0.0899,
+      "step": 23930
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.7011207970112075e-05,
+      "loss": 0.2389,
+      "step": 23931
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.6998754669987544e-05,
+      "loss": 0.1597,
+      "step": 23932
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.698630136986301e-05,
+      "loss": 0.1097,
+      "step": 23933
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.697384806973848e-05,
+      "loss": 0.0996,
+      "step": 23934
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.6961394769613942e-05,
+      "loss": 0.0461,
+      "step": 23935
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.694894146948941e-05,
+      "loss": 0.1017,
+      "step": 23936
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 2.693648816936488e-05,
+      "loss": 0.1301,
+      "step": 23937
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6924034869240347e-05,
+      "loss": 0.2043,
+      "step": 23938
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6911581569115813e-05,
+      "loss": 0.1469,
+      "step": 23939
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.689912826899128e-05,
+      "loss": 0.0912,
+      "step": 23940
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6886674968866748e-05,
+      "loss": 0.1148,
+      "step": 23941
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6874221668742217e-05,
+      "loss": 0.2529,
+      "step": 23942
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.686176836861768e-05,
+      "loss": 0.1276,
+      "step": 23943
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.684931506849315e-05,
+      "loss": 0.1472,
+      "step": 23944
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6836861768368615e-05,
+      "loss": 0.1224,
+      "step": 23945
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6824408468244084e-05,
+      "loss": 0.102,
+      "step": 23946
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6811955168119547e-05,
+      "loss": 0.1261,
+      "step": 23947
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6799501867995016e-05,
+      "loss": 0.128,
+      "step": 23948
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6787048567870482e-05,
+      "loss": 0.1971,
+      "step": 23949
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.677459526774595e-05,
+      "loss": 0.165,
+      "step": 23950
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6762141967621414e-05,
+      "loss": 0.1303,
+      "step": 23951
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6749688667496883e-05,
+      "loss": 0.1401,
+      "step": 23952
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6737235367372353e-05,
+      "loss": 0.1062,
+      "step": 23953
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.672478206724782e-05,
+      "loss": 0.0858,
+      "step": 23954
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6712328767123285e-05,
+      "loss": 0.1308,
+      "step": 23955
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.669987546699875e-05,
+      "loss": 0.0503,
+      "step": 23956
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.668742216687422e-05,
+      "loss": 0.145,
+      "step": 23957
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.667496886674969e-05,
+      "loss": 0.1634,
+      "step": 23958
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6662515566625152e-05,
+      "loss": 0.1318,
+      "step": 23959
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.665006226650062e-05,
+      "loss": 0.1388,
+      "step": 23960
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6637608966376087e-05,
+      "loss": 0.0742,
+      "step": 23961
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.6625155666251556e-05,
+      "loss": 0.0554,
+      "step": 23962
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 2.661270236612702e-05,
+      "loss": 0.1461,
+      "step": 23963
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.660024906600249e-05,
+      "loss": 0.0929,
+      "step": 23964
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6587795765877958e-05,
+      "loss": 0.0814,
+      "step": 23965
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6575342465753424e-05,
+      "loss": 0.1419,
+      "step": 23966
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.656288916562889e-05,
+      "loss": 0.1119,
+      "step": 23967
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6550435865504356e-05,
+      "loss": 0.1433,
+      "step": 23968
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6537982565379825e-05,
+      "loss": 0.1707,
+      "step": 23969
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.652552926525529e-05,
+      "loss": 0.1805,
+      "step": 23970
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6513075965130757e-05,
+      "loss": 0.1542,
+      "step": 23971
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6500622665006223e-05,
+      "loss": 0.1107,
+      "step": 23972
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6488169364881692e-05,
+      "loss": 0.0922,
+      "step": 23973
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.647571606475716e-05,
+      "loss": 0.0825,
+      "step": 23974
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6463262764632624e-05,
+      "loss": 0.1357,
+      "step": 23975
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6450809464508093e-05,
+      "loss": 0.1349,
+      "step": 23976
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.643835616438356e-05,
+      "loss": 0.1658,
+      "step": 23977
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6425902864259025e-05,
+      "loss": 0.0725,
+      "step": 23978
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.641344956413449e-05,
+      "loss": 0.068,
+      "step": 23979
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.640099626400996e-05,
+      "loss": 0.171,
+      "step": 23980
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.638854296388543e-05,
+      "loss": 0.1268,
+      "step": 23981
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6376089663760892e-05,
+      "loss": 0.1589,
+      "step": 23982
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.636363636363636e-05,
+      "loss": 0.0872,
+      "step": 23983
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6351183063511828e-05,
+      "loss": 0.1226,
+      "step": 23984
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6338729763387297e-05,
+      "loss": 0.0832,
+      "step": 23985
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.632627646326276e-05,
+      "loss": 0.1075,
+      "step": 23986
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.631382316313823e-05,
+      "loss": 0.0644,
+      "step": 23987
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6301369863013698e-05,
+      "loss": 0.1158,
+      "step": 23988
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 2.6288916562889164e-05,
+      "loss": 0.0762,
+      "step": 23989
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.627646326276463e-05,
+      "loss": 0.101,
+      "step": 23990
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6264009962640096e-05,
+      "loss": 0.063,
+      "step": 23991
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6251556662515565e-05,
+      "loss": 0.0715,
+      "step": 23992
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.623910336239103e-05,
+      "loss": 0.0877,
+      "step": 23993
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6226650062266497e-05,
+      "loss": 0.1577,
+      "step": 23994
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6214196762141963e-05,
+      "loss": 0.1506,
+      "step": 23995
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6201743462017432e-05,
+      "loss": 0.1656,
+      "step": 23996
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6189290161892902e-05,
+      "loss": 0.0652,
+      "step": 23997
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6176836861768364e-05,
+      "loss": 0.0676,
+      "step": 23998
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6164383561643834e-05,
+      "loss": 0.1398,
+      "step": 23999
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.61519302615193e-05,
+      "loss": 0.2286,
+      "step": 24000
+    },
+    {
+      "epoch": 9.2,
+      "eval_cer": 0.10085256647872486,
+      "eval_loss": 0.4148615896701813,
+      "eval_runtime": 47.775,
+      "eval_samples_per_second": 36.693,
+      "eval_steps_per_second": 4.605,
+      "eval_wer": 0.15961442729782324,
+      "step": 24000
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.613947696139477e-05,
+      "loss": 0.1232,
+      "step": 24001
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.612702366127023e-05,
+      "loss": 0.0931,
+      "step": 24002
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.61145703611457e-05,
+      "loss": 0.1322,
+      "step": 24003
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.610211706102117e-05,
+      "loss": 0.0666,
+      "step": 24004
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6089663760896636e-05,
+      "loss": 0.048,
+      "step": 24005
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6077210460772102e-05,
+      "loss": 0.2178,
+      "step": 24006
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6064757160647568e-05,
+      "loss": 0.1199,
+      "step": 24007
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6052303860523037e-05,
+      "loss": 0.0931,
+      "step": 24008
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6039850560398503e-05,
+      "loss": 0.1495,
+      "step": 24009
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.602739726027397e-05,
+      "loss": 0.0645,
+      "step": 24010
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.601494396014944e-05,
+      "loss": 0.1008,
+      "step": 24011
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.6002490660024905e-05,
+      "loss": 0.0614,
+      "step": 24012
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.5990037359900374e-05,
+      "loss": 0.1643,
+      "step": 24013
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.5977584059775836e-05,
+      "loss": 0.1441,
+      "step": 24014
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.5965130759651306e-05,
+      "loss": 0.1695,
+      "step": 24015
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5952677459526772e-05,
+      "loss": 0.1079,
+      "step": 24016
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.594022415940224e-05,
+      "loss": 0.1157,
+      "step": 24017
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5927770859277704e-05,
+      "loss": 0.0667,
+      "step": 24018
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5915317559153173e-05,
+      "loss": 0.1916,
+      "step": 24019
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5902864259028642e-05,
+      "loss": 0.164,
+      "step": 24020
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5890410958904108e-05,
+      "loss": 0.1477,
+      "step": 24021
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5877957658779574e-05,
+      "loss": 0.0722,
+      "step": 24022
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.586550435865504e-05,
+      "loss": 0.0865,
+      "step": 24023
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.585305105853051e-05,
+      "loss": 0.1094,
+      "step": 24024
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.584059775840598e-05,
+      "loss": 0.1566,
+      "step": 24025
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.582814445828144e-05,
+      "loss": 0.1155,
+      "step": 24026
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.581569115815691e-05,
+      "loss": 0.1548,
+      "step": 24027
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5803237858032377e-05,
+      "loss": 0.1093,
+      "step": 24028
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5790784557907846e-05,
+      "loss": 0.1222,
+      "step": 24029
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.577833125778331e-05,
+      "loss": 0.1111,
+      "step": 24030
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5765877957658778e-05,
+      "loss": 0.1307,
+      "step": 24031
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5753424657534244e-05,
+      "loss": 0.156,
+      "step": 24032
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.574097135740971e-05,
+      "loss": 0.1989,
+      "step": 24033
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5728518057285176e-05,
+      "loss": 0.1622,
+      "step": 24034
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5716064757160645e-05,
+      "loss": 0.0642,
+      "step": 24035
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5703611457036114e-05,
+      "loss": 0.1658,
+      "step": 24036
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5691158156911577e-05,
+      "loss": 0.0786,
+      "step": 24037
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5678704856787046e-05,
+      "loss": 0.1556,
+      "step": 24038
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5666251556662512e-05,
+      "loss": 0.1617,
+      "step": 24039
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.565379825653798e-05,
+      "loss": 0.1538,
+      "step": 24040
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 2.5641344956413444e-05,
+      "loss": 0.1198,
+      "step": 24041
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5628891656288913e-05,
+      "loss": 0.1334,
+      "step": 24042
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5616438356164383e-05,
+      "loss": 0.0422,
+      "step": 24043
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.560398505603985e-05,
+      "loss": 0.1719,
+      "step": 24044
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5591531755915315e-05,
+      "loss": 0.1158,
+      "step": 24045
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.557907845579078e-05,
+      "loss": 0.1367,
+      "step": 24046
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.556662515566625e-05,
+      "loss": 0.0568,
+      "step": 24047
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.555417185554172e-05,
+      "loss": 0.0765,
+      "step": 24048
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5541718555417182e-05,
+      "loss": 0.131,
+      "step": 24049
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.552926525529265e-05,
+      "loss": 0.1113,
+      "step": 24050
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5516811955168117e-05,
+      "loss": 0.1643,
+      "step": 24051
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5504358655043586e-05,
+      "loss": 0.1317,
+      "step": 24052
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.549190535491905e-05,
+      "loss": 0.1186,
+      "step": 24053
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5479452054794518e-05,
+      "loss": 0.0889,
+      "step": 24054
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5466998754669984e-05,
+      "loss": 0.155,
+      "step": 24055
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5454545454545454e-05,
+      "loss": 0.1602,
+      "step": 24056
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5442092154420916e-05,
+      "loss": 0.147,
+      "step": 24057
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5429638854296385e-05,
+      "loss": 0.0937,
+      "step": 24058
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5417185554171855e-05,
+      "loss": 0.1033,
+      "step": 24059
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.540473225404732e-05,
+      "loss": 0.0944,
+      "step": 24060
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5392278953922787e-05,
+      "loss": 0.1146,
+      "step": 24061
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5379825653798253e-05,
+      "loss": 0.1511,
+      "step": 24062
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5367372353673722e-05,
+      "loss": 0.2739,
+      "step": 24063
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.535491905354919e-05,
+      "loss": 0.1185,
+      "step": 24064
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5342465753424654e-05,
+      "loss": 0.0748,
+      "step": 24065
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.5330012453300123e-05,
+      "loss": 0.0813,
+      "step": 24066
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.531755915317559e-05,
+      "loss": 0.1156,
+      "step": 24067
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.530510585305106e-05,
+      "loss": 0.1202,
+      "step": 24068
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.529265255292652e-05,
+      "loss": 0.109,
+      "step": 24069
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.528019925280199e-05,
+      "loss": 0.1287,
+      "step": 24070
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.526774595267746e-05,
+      "loss": 0.1672,
+      "step": 24071
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5255292652552926e-05,
+      "loss": 0.1188,
+      "step": 24072
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.524283935242839e-05,
+      "loss": 0.0786,
+      "step": 24073
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5230386052303857e-05,
+      "loss": 0.0576,
+      "step": 24074
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5217932752179327e-05,
+      "loss": 0.1962,
+      "step": 24075
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5205479452054793e-05,
+      "loss": 0.114,
+      "step": 24076
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.519302615193026e-05,
+      "loss": 0.1874,
+      "step": 24077
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5180572851805725e-05,
+      "loss": 0.1111,
+      "step": 24078
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5168119551681194e-05,
+      "loss": 0.0629,
+      "step": 24079
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5155666251556663e-05,
+      "loss": 0.0989,
+      "step": 24080
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5143212951432126e-05,
+      "loss": 0.0962,
+      "step": 24081
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5130759651307595e-05,
+      "loss": 0.1609,
+      "step": 24082
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.511830635118306e-05,
+      "loss": 0.1143,
+      "step": 24083
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5105853051058527e-05,
+      "loss": 0.0807,
+      "step": 24084
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5093399750933993e-05,
+      "loss": 0.0638,
+      "step": 24085
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5080946450809462e-05,
+      "loss": 0.1314,
+      "step": 24086
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5068493150684932e-05,
+      "loss": 0.1492,
+      "step": 24087
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5056039850560394e-05,
+      "loss": 0.1602,
+      "step": 24088
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.5043586550435864e-05,
+      "loss": 0.1633,
+      "step": 24089
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.503113325031133e-05,
+      "loss": 0.1801,
+      "step": 24090
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.50186799501868e-05,
+      "loss": 0.0662,
+      "step": 24091
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.500622665006226e-05,
+      "loss": 0.0634,
+      "step": 24092
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.499377334993773e-05,
+      "loss": 0.0829,
+      "step": 24093
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 2.49813200498132e-05,
+      "loss": 0.1662,
+      "step": 24094
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4968866749688666e-05,
+      "loss": 0.1341,
+      "step": 24095
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4956413449564132e-05,
+      "loss": 0.0964,
+      "step": 24096
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4943960149439598e-05,
+      "loss": 0.0867,
+      "step": 24097
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4931506849315067e-05,
+      "loss": 0.1477,
+      "step": 24098
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4919053549190533e-05,
+      "loss": 0.079,
+      "step": 24099
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4906600249066e-05,
+      "loss": 0.1325,
+      "step": 24100
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4894146948941465e-05,
+      "loss": 0.1607,
+      "step": 24101
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4881693648816934e-05,
+      "loss": 0.113,
+      "step": 24102
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4869240348692404e-05,
+      "loss": 0.1034,
+      "step": 24103
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4856787048567866e-05,
+      "loss": 0.0706,
+      "step": 24104
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4844333748443336e-05,
+      "loss": 0.0601,
+      "step": 24105
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.48318804483188e-05,
+      "loss": 0.0857,
+      "step": 24106
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.481942714819427e-05,
+      "loss": 0.1289,
+      "step": 24107
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4806973848069734e-05,
+      "loss": 0.127,
+      "step": 24108
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4794520547945203e-05,
+      "loss": 0.0931,
+      "step": 24109
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4782067247820672e-05,
+      "loss": 0.0889,
+      "step": 24110
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4769613947696138e-05,
+      "loss": 0.1429,
+      "step": 24111
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4757160647571604e-05,
+      "loss": 0.0854,
+      "step": 24112
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.474470734744707e-05,
+      "loss": 0.2196,
+      "step": 24113
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.473225404732254e-05,
+      "loss": 0.0983,
+      "step": 24114
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4719800747198005e-05,
+      "loss": 0.0627,
+      "step": 24115
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.470734744707347e-05,
+      "loss": 0.1422,
+      "step": 24116
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.469489414694894e-05,
+      "loss": 0.0918,
+      "step": 24117
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4682440846824407e-05,
+      "loss": 0.0682,
+      "step": 24118
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.4669987546699876e-05,
+      "loss": 0.1579,
+      "step": 24119
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.465753424657534e-05,
+      "loss": 0.1225,
+      "step": 24120
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4645080946450808e-05,
+      "loss": 0.0938,
+      "step": 24121
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4632627646326274e-05,
+      "loss": 0.1278,
+      "step": 24122
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4620174346201743e-05,
+      "loss": 0.1898,
+      "step": 24123
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4607721046077206e-05,
+      "loss": 0.1205,
+      "step": 24124
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4595267745952675e-05,
+      "loss": 0.1969,
+      "step": 24125
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4582814445828144e-05,
+      "loss": 0.1549,
+      "step": 24126
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.457036114570361e-05,
+      "loss": 0.102,
+      "step": 24127
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4557907845579076e-05,
+      "loss": 0.1454,
+      "step": 24128
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4545454545454542e-05,
+      "loss": 0.0667,
+      "step": 24129
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.453300124533001e-05,
+      "loss": 0.1031,
+      "step": 24130
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.452054794520548e-05,
+      "loss": 0.0819,
+      "step": 24131
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4508094645080943e-05,
+      "loss": 0.1274,
+      "step": 24132
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4495641344956413e-05,
+      "loss": 0.1201,
+      "step": 24133
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.448318804483188e-05,
+      "loss": 0.0718,
+      "step": 24134
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4470734744707348e-05,
+      "loss": 0.134,
+      "step": 24135
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.445828144458281e-05,
+      "loss": 0.1301,
+      "step": 24136
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.444582814445828e-05,
+      "loss": 0.0736,
+      "step": 24137
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4433374844333746e-05,
+      "loss": 0.1921,
+      "step": 24138
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.442092154420921e-05,
+      "loss": 0.2367,
+      "step": 24139
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.440846824408468e-05,
+      "loss": 0.1092,
+      "step": 24140
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4396014943960147e-05,
+      "loss": 0.1229,
+      "step": 24141
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4383561643835616e-05,
+      "loss": 0.0871,
+      "step": 24142
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.437110834371108e-05,
+      "loss": 0.0976,
+      "step": 24143
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4358655043586548e-05,
+      "loss": 0.2052,
+      "step": 24144
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4346201743462014e-05,
+      "loss": 0.1309,
+      "step": 24145
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.4333748443337483e-05,
+      "loss": 0.1234,
+      "step": 24146
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4321295143212946e-05,
+      "loss": 0.0842,
+      "step": 24147
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4308841843088415e-05,
+      "loss": 0.0757,
+      "step": 24148
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4296388542963885e-05,
+      "loss": 0.1265,
+      "step": 24149
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.428393524283935e-05,
+      "loss": 0.213,
+      "step": 24150
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4271481942714817e-05,
+      "loss": 0.1475,
+      "step": 24151
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4259028642590283e-05,
+      "loss": 0.1519,
+      "step": 24152
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4246575342465752e-05,
+      "loss": 0.158,
+      "step": 24153
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.423412204234122e-05,
+      "loss": 0.187,
+      "step": 24154
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4221668742216684e-05,
+      "loss": 0.2167,
+      "step": 24155
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4209215442092153e-05,
+      "loss": 0.1085,
+      "step": 24156
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.419676214196762e-05,
+      "loss": 0.1522,
+      "step": 24157
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.418430884184309e-05,
+      "loss": 0.1478,
+      "step": 24158
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.417185554171855e-05,
+      "loss": 0.0945,
+      "step": 24159
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.415940224159402e-05,
+      "loss": 0.1016,
+      "step": 24160
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4146948941469486e-05,
+      "loss": 0.1724,
+      "step": 24161
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4134495641344956e-05,
+      "loss": 0.1312,
+      "step": 24162
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4122042341220418e-05,
+      "loss": 0.1328,
+      "step": 24163
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4109589041095887e-05,
+      "loss": 0.1425,
+      "step": 24164
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4097135740971357e-05,
+      "loss": 0.0956,
+      "step": 24165
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4084682440846823e-05,
+      "loss": 0.1298,
+      "step": 24166
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.407222914072229e-05,
+      "loss": 0.121,
+      "step": 24167
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4059775840597755e-05,
+      "loss": 0.105,
+      "step": 24168
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4047322540473224e-05,
+      "loss": 0.134,
+      "step": 24169
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4034869240348693e-05,
+      "loss": 0.1538,
+      "step": 24170
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4022415940224156e-05,
+      "loss": 0.0879,
+      "step": 24171
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.4009962640099625e-05,
+      "loss": 0.0876,
+      "step": 24172
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.399750933997509e-05,
+      "loss": 0.141,
+      "step": 24173
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.398505603985056e-05,
+      "loss": 0.1288,
+      "step": 24174
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3972602739726023e-05,
+      "loss": 0.2202,
+      "step": 24175
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3960149439601492e-05,
+      "loss": 0.1769,
+      "step": 24176
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.394769613947696e-05,
+      "loss": 0.1595,
+      "step": 24177
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3935242839352428e-05,
+      "loss": 0.1081,
+      "step": 24178
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3922789539227894e-05,
+      "loss": 0.1313,
+      "step": 24179
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.391033623910336e-05,
+      "loss": 0.0873,
+      "step": 24180
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.389788293897883e-05,
+      "loss": 0.167,
+      "step": 24181
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3885429638854295e-05,
+      "loss": 0.1397,
+      "step": 24182
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.387297633872976e-05,
+      "loss": 0.1353,
+      "step": 24183
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3860523038605227e-05,
+      "loss": 0.135,
+      "step": 24184
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3848069738480696e-05,
+      "loss": 0.1282,
+      "step": 24185
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3835616438356165e-05,
+      "loss": 0.0854,
+      "step": 24186
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3823163138231628e-05,
+      "loss": 0.0977,
+      "step": 24187
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3810709838107097e-05,
+      "loss": 0.2289,
+      "step": 24188
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3798256537982563e-05,
+      "loss": 0.2561,
+      "step": 24189
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.378580323785803e-05,
+      "loss": 0.1884,
+      "step": 24190
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3773349937733495e-05,
+      "loss": 0.0928,
+      "step": 24191
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3760896637608964e-05,
+      "loss": 0.077,
+      "step": 24192
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3748443337484434e-05,
+      "loss": 0.101,
+      "step": 24193
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3735990037359896e-05,
+      "loss": 0.1561,
+      "step": 24194
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3723536737235366e-05,
+      "loss": 0.1492,
+      "step": 24195
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.371108343711083e-05,
+      "loss": 0.0987,
+      "step": 24196
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.36986301369863e-05,
+      "loss": 0.1121,
+      "step": 24197
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.3686176836861763e-05,
+      "loss": 0.1539,
+      "step": 24198
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3673723536737233e-05,
+      "loss": 0.0854,
+      "step": 24199
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3661270236612702e-05,
+      "loss": 0.2174,
+      "step": 24200
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3648816936488168e-05,
+      "loss": 0.2029,
+      "step": 24201
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3636363636363634e-05,
+      "loss": 0.1242,
+      "step": 24202
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.36239103362391e-05,
+      "loss": 0.1177,
+      "step": 24203
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.361145703611457e-05,
+      "loss": 0.0708,
+      "step": 24204
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3599003735990035e-05,
+      "loss": 0.151,
+      "step": 24205
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.35865504358655e-05,
+      "loss": 0.1999,
+      "step": 24206
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3574097135740967e-05,
+      "loss": 0.1957,
+      "step": 24207
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3561643835616436e-05,
+      "loss": 0.1689,
+      "step": 24208
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3549190535491906e-05,
+      "loss": 0.1382,
+      "step": 24209
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.353673723536737e-05,
+      "loss": 0.0671,
+      "step": 24210
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3524283935242838e-05,
+      "loss": 0.065,
+      "step": 24211
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3511830635118304e-05,
+      "loss": 0.1217,
+      "step": 24212
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3499377334993773e-05,
+      "loss": 0.1582,
+      "step": 24213
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3486924034869235e-05,
+      "loss": 0.0996,
+      "step": 24214
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3474470734744705e-05,
+      "loss": 0.1158,
+      "step": 24215
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3462017434620174e-05,
+      "loss": 0.095,
+      "step": 24216
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.344956413449564e-05,
+      "loss": 0.0946,
+      "step": 24217
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3437110834371106e-05,
+      "loss": 0.1144,
+      "step": 24218
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3424657534246572e-05,
+      "loss": 0.1387,
+      "step": 24219
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.341220423412204e-05,
+      "loss": 0.0933,
+      "step": 24220
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3399750933997507e-05,
+      "loss": 0.0939,
+      "step": 24221
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3387297633872973e-05,
+      "loss": 0.0879,
+      "step": 24222
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.3374844333748443e-05,
+      "loss": 0.1056,
+      "step": 24223
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.336239103362391e-05,
+      "loss": 0.1077,
+      "step": 24224
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3349937733499378e-05,
+      "loss": 0.1449,
+      "step": 24225
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.333748443337484e-05,
+      "loss": 0.1988,
+      "step": 24226
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.332503113325031e-05,
+      "loss": 0.1129,
+      "step": 24227
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3312577833125776e-05,
+      "loss": 0.0905,
+      "step": 24228
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3300124533001245e-05,
+      "loss": 0.1591,
+      "step": 24229
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3287671232876708e-05,
+      "loss": 0.1281,
+      "step": 24230
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3275217932752177e-05,
+      "loss": 0.2298,
+      "step": 24231
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3262764632627646e-05,
+      "loss": 0.2321,
+      "step": 24232
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3250311332503112e-05,
+      "loss": 0.0872,
+      "step": 24233
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3237858032378578e-05,
+      "loss": 0.1028,
+      "step": 24234
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3225404732254044e-05,
+      "loss": 0.0743,
+      "step": 24235
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3212951432129513e-05,
+      "loss": 0.0827,
+      "step": 24236
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3200498132004983e-05,
+      "loss": 0.081,
+      "step": 24237
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3188044831880445e-05,
+      "loss": 0.2112,
+      "step": 24238
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3175591531755915e-05,
+      "loss": 0.1726,
+      "step": 24239
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.316313823163138e-05,
+      "loss": 0.1264,
+      "step": 24240
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.315068493150685e-05,
+      "loss": 0.0744,
+      "step": 24241
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3138231631382312e-05,
+      "loss": 0.1077,
+      "step": 24242
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3125778331257782e-05,
+      "loss": 0.0945,
+      "step": 24243
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3113325031133248e-05,
+      "loss": 0.2145,
+      "step": 24244
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3100871731008714e-05,
+      "loss": 0.1542,
+      "step": 24245
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3088418430884183e-05,
+      "loss": 0.0845,
+      "step": 24246
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.307596513075965e-05,
+      "loss": 0.0649,
+      "step": 24247
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.3063511830635118e-05,
+      "loss": 0.0712,
+      "step": 24248
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.305105853051058e-05,
+      "loss": 0.0985,
+      "step": 24249
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.303860523038605e-05,
+      "loss": 0.1766,
+      "step": 24250
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.3026151930261516e-05,
+      "loss": 0.0958,
+      "step": 24251
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.3013698630136985e-05,
+      "loss": 0.0975,
+      "step": 24252
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.3001245330012448e-05,
+      "loss": 0.0955,
+      "step": 24253
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2988792029887917e-05,
+      "loss": 0.1116,
+      "step": 24254
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2976338729763387e-05,
+      "loss": 0.1006,
+      "step": 24255
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2963885429638853e-05,
+      "loss": 0.1562,
+      "step": 24256
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.295143212951432e-05,
+      "loss": 0.1007,
+      "step": 24257
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2938978829389785e-05,
+      "loss": 0.1073,
+      "step": 24258
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2926525529265254e-05,
+      "loss": 0.1149,
+      "step": 24259
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2914072229140723e-05,
+      "loss": 0.1754,
+      "step": 24260
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2901618929016186e-05,
+      "loss": 0.0857,
+      "step": 24261
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2889165628891655e-05,
+      "loss": 0.1716,
+      "step": 24262
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.287671232876712e-05,
+      "loss": 0.134,
+      "step": 24263
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.286425902864259e-05,
+      "loss": 0.1234,
+      "step": 24264
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2851805728518053e-05,
+      "loss": 0.0929,
+      "step": 24265
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2839352428393522e-05,
+      "loss": 0.1072,
+      "step": 24266
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2826899128268988e-05,
+      "loss": 0.1759,
+      "step": 24267
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2814445828144458e-05,
+      "loss": 0.1174,
+      "step": 24268
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.280199252801992e-05,
+      "loss": 0.2083,
+      "step": 24269
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.278953922789539e-05,
+      "loss": 0.1197,
+      "step": 24270
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.277708592777086e-05,
+      "loss": 0.1271,
+      "step": 24271
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2764632627646325e-05,
+      "loss": 0.1021,
+      "step": 24272
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.275217932752179e-05,
+      "loss": 0.1705,
+      "step": 24273
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2739726027397257e-05,
+      "loss": 0.0713,
+      "step": 24274
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2727272727272726e-05,
+      "loss": 0.1236,
+      "step": 24275
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.2714819427148195e-05,
+      "loss": 0.1376,
+      "step": 24276
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2702366127023658e-05,
+      "loss": 0.1903,
+      "step": 24277
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2689912826899127e-05,
+      "loss": 0.0909,
+      "step": 24278
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2677459526774593e-05,
+      "loss": 0.0565,
+      "step": 24279
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2665006226650062e-05,
+      "loss": 0.0721,
+      "step": 24280
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2652552926525525e-05,
+      "loss": 0.1281,
+      "step": 24281
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2640099626400994e-05,
+      "loss": 0.1323,
+      "step": 24282
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2627646326276464e-05,
+      "loss": 0.094,
+      "step": 24283
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.261519302615193e-05,
+      "loss": 0.162,
+      "step": 24284
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2602739726027396e-05,
+      "loss": 0.0576,
+      "step": 24285
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.259028642590286e-05,
+      "loss": 0.08,
+      "step": 24286
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.257783312577833e-05,
+      "loss": 0.1422,
+      "step": 24287
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2565379825653797e-05,
+      "loss": 0.2521,
+      "step": 24288
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2552926525529263e-05,
+      "loss": 0.1329,
+      "step": 24289
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.254047322540473e-05,
+      "loss": 0.1628,
+      "step": 24290
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2528019925280198e-05,
+      "loss": 0.0966,
+      "step": 24291
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2515566625155667e-05,
+      "loss": 0.0793,
+      "step": 24292
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.250311332503113e-05,
+      "loss": 0.1031,
+      "step": 24293
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.24906600249066e-05,
+      "loss": 0.1422,
+      "step": 24294
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2478206724782065e-05,
+      "loss": 0.1971,
+      "step": 24295
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.246575342465753e-05,
+      "loss": 0.1324,
+      "step": 24296
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2453300124532997e-05,
+      "loss": 0.1357,
+      "step": 24297
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2440846824408466e-05,
+      "loss": 0.0757,
+      "step": 24298
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2428393524283936e-05,
+      "loss": 0.0696,
+      "step": 24299
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2415940224159398e-05,
+      "loss": 0.1207,
+      "step": 24300
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2403486924034868e-05,
+      "loss": 0.1587,
+      "step": 24301
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.2391033623910334e-05,
+      "loss": 0.1298,
+      "step": 24302
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2378580323785803e-05,
+      "loss": 0.058,
+      "step": 24303
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2366127023661265e-05,
+      "loss": 0.1472,
+      "step": 24304
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2353673723536735e-05,
+      "loss": 0.095,
+      "step": 24305
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2341220423412204e-05,
+      "loss": 0.1607,
+      "step": 24306
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.232876712328767e-05,
+      "loss": 0.1186,
+      "step": 24307
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2316313823163136e-05,
+      "loss": 0.1534,
+      "step": 24308
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2303860523038602e-05,
+      "loss": 0.1429,
+      "step": 24309
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.229140722291407e-05,
+      "loss": 0.1601,
+      "step": 24310
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2278953922789537e-05,
+      "loss": 0.0892,
+      "step": 24311
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2266500622665003e-05,
+      "loss": 0.1336,
+      "step": 24312
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.225404732254047e-05,
+      "loss": 0.1387,
+      "step": 24313
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.224159402241594e-05,
+      "loss": 0.1462,
+      "step": 24314
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2229140722291408e-05,
+      "loss": 0.1126,
+      "step": 24315
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.221668742216687e-05,
+      "loss": 0.1216,
+      "step": 24316
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.220423412204234e-05,
+      "loss": 0.1585,
+      "step": 24317
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2191780821917806e-05,
+      "loss": 0.078,
+      "step": 24318
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2179327521793275e-05,
+      "loss": 0.2163,
+      "step": 24319
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2166874221668737e-05,
+      "loss": 0.1348,
+      "step": 24320
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2154420921544207e-05,
+      "loss": 0.0946,
+      "step": 24321
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2141967621419676e-05,
+      "loss": 0.0847,
+      "step": 24322
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2129514321295142e-05,
+      "loss": 0.0679,
+      "step": 24323
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2117061021170608e-05,
+      "loss": 0.1311,
+      "step": 24324
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2104607721046074e-05,
+      "loss": 0.1166,
+      "step": 24325
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2092154420921543e-05,
+      "loss": 0.1545,
+      "step": 24326
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.207970112079701e-05,
+      "loss": 0.1021,
+      "step": 24327
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.2067247820672475e-05,
+      "loss": 0.1023,
+      "step": 24328
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.2054794520547945e-05,
+      "loss": 0.0634,
+      "step": 24329
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.204234122042341e-05,
+      "loss": 0.0872,
+      "step": 24330
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.202988792029888e-05,
+      "loss": 0.1142,
+      "step": 24331
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.2017434620174342e-05,
+      "loss": 0.1798,
+      "step": 24332
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.2004981320049812e-05,
+      "loss": 0.2958,
+      "step": 24333
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1992528019925278e-05,
+      "loss": 0.0927,
+      "step": 24334
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1980074719800747e-05,
+      "loss": 0.0859,
+      "step": 24335
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.196762141967621e-05,
+      "loss": 0.0589,
+      "step": 24336
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.195516811955168e-05,
+      "loss": 0.0748,
+      "step": 24337
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1942714819427148e-05,
+      "loss": 0.2277,
+      "step": 24338
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1930261519302614e-05,
+      "loss": 0.1065,
+      "step": 24339
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.191780821917808e-05,
+      "loss": 0.0977,
+      "step": 24340
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1905354919053546e-05,
+      "loss": 0.0826,
+      "step": 24341
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1892901618929015e-05,
+      "loss": 0.0589,
+      "step": 24342
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1880448318804485e-05,
+      "loss": 0.0545,
+      "step": 24343
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1867995018679947e-05,
+      "loss": 0.1528,
+      "step": 24344
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1855541718555417e-05,
+      "loss": 0.1863,
+      "step": 24345
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1843088418430883e-05,
+      "loss": 0.1116,
+      "step": 24346
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1830635118306352e-05,
+      "loss": 0.101,
+      "step": 24347
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1818181818181814e-05,
+      "loss": 0.1223,
+      "step": 24348
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1805728518057284e-05,
+      "loss": 0.1773,
+      "step": 24349
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.179327521793275e-05,
+      "loss": 0.0767,
+      "step": 24350
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1780821917808216e-05,
+      "loss": 0.1672,
+      "step": 24351
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1768368617683685e-05,
+      "loss": 0.162,
+      "step": 24352
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.175591531755915e-05,
+      "loss": 0.1865,
+      "step": 24353
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.174346201743462e-05,
+      "loss": 0.0969,
+      "step": 24354
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.1731008717310083e-05,
+      "loss": 0.0438,
+      "step": 24355
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1718555417185552e-05,
+      "loss": 0.1145,
+      "step": 24356
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1706102117061018e-05,
+      "loss": 0.1383,
+      "step": 24357
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1693648816936487e-05,
+      "loss": 0.1268,
+      "step": 24358
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.168119551681195e-05,
+      "loss": 0.1238,
+      "step": 24359
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.166874221668742e-05,
+      "loss": 0.098,
+      "step": 24360
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.165628891656289e-05,
+      "loss": 0.1465,
+      "step": 24361
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1643835616438355e-05,
+      "loss": 0.1159,
+      "step": 24362
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.163138231631382e-05,
+      "loss": 0.2127,
+      "step": 24363
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1618929016189286e-05,
+      "loss": 0.1527,
+      "step": 24364
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1606475716064756e-05,
+      "loss": 0.1658,
+      "step": 24365
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1594022415940225e-05,
+      "loss": 0.0477,
+      "step": 24366
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1581569115815688e-05,
+      "loss": 0.0781,
+      "step": 24367
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1569115815691157e-05,
+      "loss": 0.1003,
+      "step": 24368
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1556662515566623e-05,
+      "loss": 0.1801,
+      "step": 24369
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1544209215442092e-05,
+      "loss": 0.1737,
+      "step": 24370
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1531755915317555e-05,
+      "loss": 0.0948,
+      "step": 24371
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1519302615193024e-05,
+      "loss": 0.116,
+      "step": 24372
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.150684931506849e-05,
+      "loss": 0.1058,
+      "step": 24373
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.149439601494396e-05,
+      "loss": 0.1681,
+      "step": 24374
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1481942714819422e-05,
+      "loss": 0.1051,
+      "step": 24375
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.146948941469489e-05,
+      "loss": 0.1526,
+      "step": 24376
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.145703611457036e-05,
+      "loss": 0.1463,
+      "step": 24377
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1444582814445827e-05,
+      "loss": 0.1278,
+      "step": 24378
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1432129514321293e-05,
+      "loss": 0.0958,
+      "step": 24379
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.141967621419676e-05,
+      "loss": 0.115,
+      "step": 24380
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.1407222914072228e-05,
+      "loss": 0.1066,
+      "step": 24381
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1394769613947697e-05,
+      "loss": 0.1562,
+      "step": 24382
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.138231631382316e-05,
+      "loss": 0.1781,
+      "step": 24383
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.136986301369863e-05,
+      "loss": 0.0956,
+      "step": 24384
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1357409713574095e-05,
+      "loss": 0.0709,
+      "step": 24385
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1344956413449564e-05,
+      "loss": 0.079,
+      "step": 24386
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1332503113325027e-05,
+      "loss": 0.1191,
+      "step": 24387
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1320049813200496e-05,
+      "loss": 0.1453,
+      "step": 24388
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1307596513075966e-05,
+      "loss": 0.1412,
+      "step": 24389
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.129514321295143e-05,
+      "loss": 0.1039,
+      "step": 24390
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1282689912826897e-05,
+      "loss": 0.109,
+      "step": 24391
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1270236612702363e-05,
+      "loss": 0.0555,
+      "step": 24392
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1257783312577833e-05,
+      "loss": 0.1298,
+      "step": 24393
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.12453300124533e-05,
+      "loss": 0.2448,
+      "step": 24394
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1232876712328765e-05,
+      "loss": 0.1038,
+      "step": 24395
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.122042341220423e-05,
+      "loss": 0.2148,
+      "step": 24396
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.12079701120797e-05,
+      "loss": 0.1593,
+      "step": 24397
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.119551681195517e-05,
+      "loss": 0.0644,
+      "step": 24398
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1183063511830632e-05,
+      "loss": 0.0975,
+      "step": 24399
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.11706102117061e-05,
+      "loss": 0.2074,
+      "step": 24400
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1158156911581567e-05,
+      "loss": 0.1389,
+      "step": 24401
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1145703611457036e-05,
+      "loss": 0.1695,
+      "step": 24402
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.11332503113325e-05,
+      "loss": 0.0789,
+      "step": 24403
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.112079701120797e-05,
+      "loss": 0.1246,
+      "step": 24404
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.1108343711083438e-05,
+      "loss": 0.0776,
+      "step": 24405
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.10958904109589e-05,
+      "loss": 0.135,
+      "step": 24406
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.108343711083437e-05,
+      "loss": 0.1585,
+      "step": 24407
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.1070983810709835e-05,
+      "loss": 0.1193,
+      "step": 24408
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.1058530510585305e-05,
+      "loss": 0.2023,
+      "step": 24409
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.1046077210460767e-05,
+      "loss": 0.0956,
+      "step": 24410
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.1033623910336237e-05,
+      "loss": 0.1924,
+      "step": 24411
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.1021170610211706e-05,
+      "loss": 0.1066,
+      "step": 24412
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.1008717310087172e-05,
+      "loss": 0.2778,
+      "step": 24413
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.0996264009962638e-05,
+      "loss": 0.1311,
+      "step": 24414
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.0983810709838104e-05,
+      "loss": 0.1067,
+      "step": 24415
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.0971357409713573e-05,
+      "loss": 0.0993,
+      "step": 24416
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.095890410958904e-05,
+      "loss": 0.149,
+      "step": 24417
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.0946450809464505e-05,
+      "loss": 0.0801,
+      "step": 24418
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.093399750933997e-05,
+      "loss": 0.1019,
+      "step": 24419
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.092154420921544e-05,
+      "loss": 0.1303,
+      "step": 24420
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.090909090909091e-05,
+      "loss": 0.1422,
+      "step": 24421
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.0896637608966372e-05,
+      "loss": 0.0589,
+      "step": 24422
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.088418430884184e-05,
+      "loss": 0.0525,
+      "step": 24423
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.0871731008717308e-05,
+      "loss": 0.1345,
+      "step": 24424
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.0859277708592777e-05,
+      "loss": 0.1761,
+      "step": 24425
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.084682440846824e-05,
+      "loss": 0.0842,
+      "step": 24426
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.083437110834371e-05,
+      "loss": 0.0864,
+      "step": 24427
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.0821917808219178e-05,
+      "loss": 0.2013,
+      "step": 24428
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.0809464508094644e-05,
+      "loss": 0.0648,
+      "step": 24429
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.079701120797011e-05,
+      "loss": 0.0542,
+      "step": 24430
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.0784557907845576e-05,
+      "loss": 0.1498,
+      "step": 24431
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.0772104607721045e-05,
+      "loss": 0.1459,
+      "step": 24432
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.075965130759651e-05,
+      "loss": 0.091,
+      "step": 24433
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0747198007471977e-05,
+      "loss": 0.1549,
+      "step": 24434
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0734744707347447e-05,
+      "loss": 0.1291,
+      "step": 24435
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0722291407222912e-05,
+      "loss": 0.131,
+      "step": 24436
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0709838107098382e-05,
+      "loss": 0.0935,
+      "step": 24437
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0697384806973844e-05,
+      "loss": 0.2067,
+      "step": 24438
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0684931506849314e-05,
+      "loss": 0.196,
+      "step": 24439
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.067247820672478e-05,
+      "loss": 0.0962,
+      "step": 24440
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.066002490660025e-05,
+      "loss": 0.1226,
+      "step": 24441
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.064757160647571e-05,
+      "loss": 0.1072,
+      "step": 24442
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.063511830635118e-05,
+      "loss": 0.0883,
+      "step": 24443
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.062266500622665e-05,
+      "loss": 0.1556,
+      "step": 24444
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0610211706102116e-05,
+      "loss": 0.1176,
+      "step": 24445
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0597758405977582e-05,
+      "loss": 0.0961,
+      "step": 24446
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0585305105853048e-05,
+      "loss": 0.0707,
+      "step": 24447
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0572851805728517e-05,
+      "loss": 0.1611,
+      "step": 24448
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0560398505603987e-05,
+      "loss": 0.1002,
+      "step": 24449
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.054794520547945e-05,
+      "loss": 0.2102,
+      "step": 24450
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.053549190535492e-05,
+      "loss": 0.1663,
+      "step": 24451
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0523038605230385e-05,
+      "loss": 0.2308,
+      "step": 24452
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0510585305105854e-05,
+      "loss": 0.1081,
+      "step": 24453
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0498132004981316e-05,
+      "loss": 0.1586,
+      "step": 24454
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0485678704856786e-05,
+      "loss": 0.0608,
+      "step": 24455
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.047322540473225e-05,
+      "loss": 0.1519,
+      "step": 24456
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0460772104607718e-05,
+      "loss": 0.1555,
+      "step": 24457
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0448318804483187e-05,
+      "loss": 0.1235,
+      "step": 24458
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0435865504358653e-05,
+      "loss": 0.101,
+      "step": 24459
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0423412204234122e-05,
+      "loss": 0.1446,
+      "step": 24460
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0410958904109585e-05,
+      "loss": 0.1209,
+      "step": 24461
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0398505603985054e-05,
+      "loss": 0.2769,
+      "step": 24462
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.038605230386052e-05,
+      "loss": 0.1554,
+      "step": 24463
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.037359900373599e-05,
+      "loss": 0.1406,
+      "step": 24464
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0361145703611452e-05,
+      "loss": 0.1539,
+      "step": 24465
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.034869240348692e-05,
+      "loss": 0.1019,
+      "step": 24466
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.033623910336239e-05,
+      "loss": 0.1154,
+      "step": 24467
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0323785803237857e-05,
+      "loss": 0.1154,
+      "step": 24468
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0311332503113323e-05,
+      "loss": 0.1425,
+      "step": 24469
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.029887920298879e-05,
+      "loss": 0.135,
+      "step": 24470
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0286425902864258e-05,
+      "loss": 0.1018,
+      "step": 24471
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0273972602739727e-05,
+      "loss": 0.1133,
+      "step": 24472
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.026151930261519e-05,
+      "loss": 0.095,
+      "step": 24473
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.024906600249066e-05,
+      "loss": 0.0687,
+      "step": 24474
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0236612702366125e-05,
+      "loss": 0.1551,
+      "step": 24475
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0224159402241594e-05,
+      "loss": 0.1776,
+      "step": 24476
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0211706102117057e-05,
+      "loss": 0.0773,
+      "step": 24477
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0199252801992526e-05,
+      "loss": 0.0657,
+      "step": 24478
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0186799501867992e-05,
+      "loss": 0.1154,
+      "step": 24479
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.017434620174346e-05,
+      "loss": 0.0683,
+      "step": 24480
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0161892901618924e-05,
+      "loss": 0.074,
+      "step": 24481
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0149439601494393e-05,
+      "loss": 0.1285,
+      "step": 24482
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0136986301369863e-05,
+      "loss": 0.1074,
+      "step": 24483
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.012453300124533e-05,
+      "loss": 0.1621,
+      "step": 24484
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0112079701120795e-05,
+      "loss": 0.1075,
+      "step": 24485
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.009962640099626e-05,
+      "loss": 0.0661,
+      "step": 24486
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.008717310087173e-05,
+      "loss": 0.0695,
+      "step": 24487
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.00747198007472e-05,
+      "loss": 0.2359,
+      "step": 24488
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.0062266500622662e-05,
+      "loss": 0.1363,
+      "step": 24489
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.004981320049813e-05,
+      "loss": 0.0954,
+      "step": 24490
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.0037359900373597e-05,
+      "loss": 0.094,
+      "step": 24491
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.0024906600249066e-05,
+      "loss": 0.0733,
+      "step": 24492
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 2.001245330012453e-05,
+      "loss": 0.1433,
+      "step": 24493
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 0.1947,
+      "step": 24494
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.9987546699875468e-05,
+      "loss": 0.1988,
+      "step": 24495
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.9975093399750934e-05,
+      "loss": 0.1975,
+      "step": 24496
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.99626400996264e-05,
+      "loss": 0.0972,
+      "step": 24497
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.9950186799501865e-05,
+      "loss": 0.0818,
+      "step": 24498
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.9937733499377335e-05,
+      "loss": 0.0791,
+      "step": 24499
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.99252801992528e-05,
+      "loss": 0.1284,
+      "step": 24500
+    },
+    {
+      "epoch": 9.39,
+      "eval_cer": 0.1002360653179078,
+      "eval_loss": 0.4250880181789398,
+      "eval_runtime": 47.6367,
+      "eval_samples_per_second": 36.799,
+      "eval_steps_per_second": 4.618,
+      "eval_wer": 0.15723853916821287,
+      "step": 24500
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.9912826899128267e-05,
+      "loss": 0.1303,
+      "step": 24501
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.9900373599003733e-05,
+      "loss": 0.0938,
+      "step": 24502
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.9887920298879202e-05,
+      "loss": 0.1085,
+      "step": 24503
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.987546699875467e-05,
+      "loss": 0.0763,
+      "step": 24504
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.9863013698630134e-05,
+      "loss": 0.0941,
+      "step": 24505
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.9850560398505603e-05,
+      "loss": 0.1228,
+      "step": 24506
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.983810709838107e-05,
+      "loss": 0.12,
+      "step": 24507
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.982565379825654e-05,
+      "loss": 0.2263,
+      "step": 24508
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.9813200498132e-05,
+      "loss": 0.085,
+      "step": 24509
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.980074719800747e-05,
+      "loss": 0.1597,
+      "step": 24510
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.978829389788294e-05,
+      "loss": 0.0614,
+      "step": 24511
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9775840597758402e-05,
+      "loss": 0.1672,
+      "step": 24512
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.976338729763387e-05,
+      "loss": 0.2316,
+      "step": 24513
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9750933997509337e-05,
+      "loss": 0.0892,
+      "step": 24514
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9738480697384807e-05,
+      "loss": 0.1096,
+      "step": 24515
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.972602739726027e-05,
+      "loss": 0.0726,
+      "step": 24516
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.971357409713574e-05,
+      "loss": 0.0977,
+      "step": 24517
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9701120797011208e-05,
+      "loss": 0.1568,
+      "step": 24518
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9688667496886674e-05,
+      "loss": 0.1229,
+      "step": 24519
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.967621419676214e-05,
+      "loss": 0.1217,
+      "step": 24520
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9663760896637606e-05,
+      "loss": 0.1032,
+      "step": 24521
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9651307596513075e-05,
+      "loss": 0.0813,
+      "step": 24522
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.963885429638854e-05,
+      "loss": 0.1496,
+      "step": 24523
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9626400996264007e-05,
+      "loss": 0.0694,
+      "step": 24524
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9613947696139473e-05,
+      "loss": 0.292,
+      "step": 24525
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9601494396014942e-05,
+      "loss": 0.1202,
+      "step": 24526
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9589041095890412e-05,
+      "loss": 0.1183,
+      "step": 24527
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9576587795765874e-05,
+      "loss": 0.0816,
+      "step": 24528
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9564134495641344e-05,
+      "loss": 0.0812,
+      "step": 24529
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.955168119551681e-05,
+      "loss": 0.0793,
+      "step": 24530
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.953922789539228e-05,
+      "loss": 0.1437,
+      "step": 24531
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.952677459526774e-05,
+      "loss": 0.1772,
+      "step": 24532
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.951432129514321e-05,
+      "loss": 0.1123,
+      "step": 24533
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.950186799501868e-05,
+      "loss": 0.0709,
+      "step": 24534
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9489414694894146e-05,
+      "loss": 0.0774,
+      "step": 24535
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9476961394769612e-05,
+      "loss": 0.1456,
+      "step": 24536
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.9464508094645078e-05,
+      "loss": 0.1296,
+      "step": 24537
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9452054794520547e-05,
+      "loss": 0.2491,
+      "step": 24538
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9439601494396013e-05,
+      "loss": 0.1333,
+      "step": 24539
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.942714819427148e-05,
+      "loss": 0.0742,
+      "step": 24540
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.941469489414695e-05,
+      "loss": 0.0936,
+      "step": 24541
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9402241594022414e-05,
+      "loss": 0.064,
+      "step": 24542
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9389788293897884e-05,
+      "loss": 0.0745,
+      "step": 24543
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9377334993773346e-05,
+      "loss": 0.2527,
+      "step": 24544
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9364881693648816e-05,
+      "loss": 0.1587,
+      "step": 24545
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.935242839352428e-05,
+      "loss": 0.0956,
+      "step": 24546
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.933997509339975e-05,
+      "loss": 0.1359,
+      "step": 24547
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9327521793275213e-05,
+      "loss": 0.0833,
+      "step": 24548
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9315068493150683e-05,
+      "loss": 0.1642,
+      "step": 24549
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9302615193026152e-05,
+      "loss": 0.2063,
+      "step": 24550
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9290161892901618e-05,
+      "loss": 0.1462,
+      "step": 24551
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9277708592777084e-05,
+      "loss": 0.1025,
+      "step": 24552
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.926525529265255e-05,
+      "loss": 0.087,
+      "step": 24553
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.925280199252802e-05,
+      "loss": 0.0908,
+      "step": 24554
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.924034869240349e-05,
+      "loss": 0.1447,
+      "step": 24555
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.922789539227895e-05,
+      "loss": 0.1098,
+      "step": 24556
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.921544209215442e-05,
+      "loss": 0.1295,
+      "step": 24557
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9202988792029886e-05,
+      "loss": 0.1033,
+      "step": 24558
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9190535491905356e-05,
+      "loss": 0.2226,
+      "step": 24559
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.917808219178082e-05,
+      "loss": 0.1375,
+      "step": 24560
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9165628891656288e-05,
+      "loss": 0.0904,
+      "step": 24561
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.9153175591531754e-05,
+      "loss": 0.1604,
+      "step": 24562
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.914072229140722e-05,
+      "loss": 0.2508,
+      "step": 24563
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.912826899128269e-05,
+      "loss": 0.1529,
+      "step": 24564
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.9115815691158155e-05,
+      "loss": 0.1482,
+      "step": 24565
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.9103362391033624e-05,
+      "loss": 0.0797,
+      "step": 24566
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.9090909090909087e-05,
+      "loss": 0.0963,
+      "step": 24567
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.9078455790784556e-05,
+      "loss": 0.0905,
+      "step": 24568
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.9066002490660022e-05,
+      "loss": 0.1269,
+      "step": 24569
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.905354919053549e-05,
+      "loss": 0.24,
+      "step": 24570
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.9041095890410954e-05,
+      "loss": 0.1214,
+      "step": 24571
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.9028642590286423e-05,
+      "loss": 0.2464,
+      "step": 24572
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.9016189290161893e-05,
+      "loss": 0.1032,
+      "step": 24573
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.900373599003736e-05,
+      "loss": 0.1102,
+      "step": 24574
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8991282689912824e-05,
+      "loss": 0.1469,
+      "step": 24575
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.897882938978829e-05,
+      "loss": 0.1144,
+      "step": 24576
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.896637608966376e-05,
+      "loss": 0.0932,
+      "step": 24577
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.895392278953923e-05,
+      "loss": 0.0954,
+      "step": 24578
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.894146948941469e-05,
+      "loss": 0.0776,
+      "step": 24579
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.892901618929016e-05,
+      "loss": 0.0562,
+      "step": 24580
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8916562889165627e-05,
+      "loss": 0.1361,
+      "step": 24581
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8904109589041096e-05,
+      "loss": 0.1958,
+      "step": 24582
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.889165628891656e-05,
+      "loss": 0.1184,
+      "step": 24583
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8879202988792028e-05,
+      "loss": 0.1748,
+      "step": 24584
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8866749688667494e-05,
+      "loss": 0.0807,
+      "step": 24585
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8854296388542963e-05,
+      "loss": 0.1704,
+      "step": 24586
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8841843088418426e-05,
+      "loss": 0.1174,
+      "step": 24587
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8829389788293895e-05,
+      "loss": 0.1992,
+      "step": 24588
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8816936488169365e-05,
+      "loss": 0.1509,
+      "step": 24589
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.880448318804483e-05,
+      "loss": 0.1531,
+      "step": 24590
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8792029887920297e-05,
+      "loss": 0.0645,
+      "step": 24591
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8779576587795763e-05,
+      "loss": 0.0926,
+      "step": 24592
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8767123287671232e-05,
+      "loss": 0.0822,
+      "step": 24593
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.87546699875467e-05,
+      "loss": 0.2012,
+      "step": 24594
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8742216687422167e-05,
+      "loss": 0.1075,
+      "step": 24595
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8729763387297633e-05,
+      "loss": 0.146,
+      "step": 24596
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.87173100871731e-05,
+      "loss": 0.113,
+      "step": 24597
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8704856787048565e-05,
+      "loss": 0.2026,
+      "step": 24598
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8692403486924034e-05,
+      "loss": 0.0808,
+      "step": 24599
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.86799501867995e-05,
+      "loss": 0.2234,
+      "step": 24600
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.866749688667497e-05,
+      "loss": 0.1048,
+      "step": 24601
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8655043586550436e-05,
+      "loss": 0.1192,
+      "step": 24602
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.86425902864259e-05,
+      "loss": 0.1641,
+      "step": 24603
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8630136986301367e-05,
+      "loss": 0.1339,
+      "step": 24604
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8617683686176837e-05,
+      "loss": 0.1807,
+      "step": 24605
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8605230386052303e-05,
+      "loss": 0.0922,
+      "step": 24606
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.859277708592777e-05,
+      "loss": 0.1224,
+      "step": 24607
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8580323785803235e-05,
+      "loss": 0.1451,
+      "step": 24608
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8567870485678704e-05,
+      "loss": 0.1018,
+      "step": 24609
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.855541718555417e-05,
+      "loss": 0.1755,
+      "step": 24610
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.854296388542964e-05,
+      "loss": 0.0779,
+      "step": 24611
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8530510585305105e-05,
+      "loss": 0.1459,
+      "step": 24612
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.851805728518057e-05,
+      "loss": 0.1354,
+      "step": 24613
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8505603985056037e-05,
+      "loss": 0.113,
+      "step": 24614
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8493150684931503e-05,
+      "loss": 0.1246,
+      "step": 24615
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8480697384806972e-05,
+      "loss": 0.2677,
+      "step": 24616
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8468244084682438e-05,
+      "loss": 0.116,
+      "step": 24617
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8455790784557908e-05,
+      "loss": 0.1347,
+      "step": 24618
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8443337484433374e-05,
+      "loss": 0.134,
+      "step": 24619
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.843088418430884e-05,
+      "loss": 0.1116,
+      "step": 24620
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8418430884184305e-05,
+      "loss": 0.1167,
+      "step": 24621
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8405977584059775e-05,
+      "loss": 0.1799,
+      "step": 24622
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.839352428393524e-05,
+      "loss": 0.0659,
+      "step": 24623
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.838107098381071e-05,
+      "loss": 0.1213,
+      "step": 24624
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8368617683686176e-05,
+      "loss": 0.2781,
+      "step": 24625
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8356164383561642e-05,
+      "loss": 0.1418,
+      "step": 24626
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8343711083437108e-05,
+      "loss": 0.108,
+      "step": 24627
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8331257783312577e-05,
+      "loss": 0.1847,
+      "step": 24628
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8318804483188043e-05,
+      "loss": 0.081,
+      "step": 24629
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.830635118306351e-05,
+      "loss": 0.076,
+      "step": 24630
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8293897882938975e-05,
+      "loss": 0.0855,
+      "step": 24631
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8281444582814444e-05,
+      "loss": 0.7035,
+      "step": 24632
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.826899128268991e-05,
+      "loss": 0.1201,
+      "step": 24633
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.825653798256538e-05,
+      "loss": 0.0841,
+      "step": 24634
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8244084682440846e-05,
+      "loss": 0.1203,
+      "step": 24635
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.823163138231631e-05,
+      "loss": 0.0929,
+      "step": 24636
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8219178082191777e-05,
+      "loss": 0.1202,
+      "step": 24637
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8206724782067247e-05,
+      "loss": 0.1985,
+      "step": 24638
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8194271481942713e-05,
+      "loss": 0.0961,
+      "step": 24639
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.0883,
+      "step": 24640
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8169364881693648e-05,
+      "loss": 0.0921,
+      "step": 24641
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8156911581569114e-05,
+      "loss": 0.1168,
+      "step": 24642
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.814445828144458e-05,
+      "loss": 0.086,
+      "step": 24643
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.813200498132005e-05,
+      "loss": 0.1763,
+      "step": 24644
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8119551681195515e-05,
+      "loss": 0.1846,
+      "step": 24645
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8107098381070985e-05,
+      "loss": 0.0859,
+      "step": 24646
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.809464508094645e-05,
+      "loss": 0.087,
+      "step": 24647
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8082191780821916e-05,
+      "loss": 0.1202,
+      "step": 24648
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8069738480697382e-05,
+      "loss": 0.0823,
+      "step": 24649
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.805728518057285e-05,
+      "loss": 0.1375,
+      "step": 24650
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8044831880448318e-05,
+      "loss": 0.128,
+      "step": 24651
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8032378580323784e-05,
+      "loss": 0.1437,
+      "step": 24652
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.801992528019925e-05,
+      "loss": 0.0931,
+      "step": 24653
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.800747198007472e-05,
+      "loss": 0.0674,
+      "step": 24654
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.7995018679950185e-05,
+      "loss": 0.1349,
+      "step": 24655
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.7982565379825654e-05,
+      "loss": 0.1344,
+      "step": 24656
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.797011207970112e-05,
+      "loss": 0.1139,
+      "step": 24657
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.7957658779576586e-05,
+      "loss": 0.1828,
+      "step": 24658
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.7945205479452052e-05,
+      "loss": 0.0817,
+      "step": 24659
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.793275217932752e-05,
+      "loss": 0.0572,
+      "step": 24660
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.7920298879202987e-05,
+      "loss": 0.1203,
+      "step": 24661
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.7907845579078457e-05,
+      "loss": 0.1108,
+      "step": 24662
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.7895392278953923e-05,
+      "loss": 0.0931,
+      "step": 24663
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.788293897882939e-05,
+      "loss": 0.1697,
+      "step": 24664
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.7870485678704854e-05,
+      "loss": 0.0998,
+      "step": 24665
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.7858032378580324e-05,
+      "loss": 0.05,
+      "step": 24666
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.784557907845579e-05,
+      "loss": 0.1318,
+      "step": 24667
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.7833125778331256e-05,
+      "loss": 0.1463,
+      "step": 24668
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.782067247820672e-05,
+      "loss": 0.181,
+      "step": 24669
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.780821917808219e-05,
+      "loss": 0.1641,
+      "step": 24670
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7795765877957657e-05,
+      "loss": 0.0965,
+      "step": 24671
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7783312577833123e-05,
+      "loss": 0.0683,
+      "step": 24672
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7770859277708592e-05,
+      "loss": 0.087,
+      "step": 24673
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7758405977584058e-05,
+      "loss": 0.0848,
+      "step": 24674
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7745952677459524e-05,
+      "loss": 0.1603,
+      "step": 24675
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.773349937733499e-05,
+      "loss": 0.0938,
+      "step": 24676
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.772104607721046e-05,
+      "loss": 0.0775,
+      "step": 24677
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7708592777085925e-05,
+      "loss": 0.1428,
+      "step": 24678
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7696139476961395e-05,
+      "loss": 0.1277,
+      "step": 24679
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.768368617683686e-05,
+      "loss": 0.1004,
+      "step": 24680
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7671232876712326e-05,
+      "loss": 0.169,
+      "step": 24681
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7658779576587792e-05,
+      "loss": 0.1701,
+      "step": 24682
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7646326276463262e-05,
+      "loss": 0.2852,
+      "step": 24683
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7633872976338728e-05,
+      "loss": 0.1364,
+      "step": 24684
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7621419676214197e-05,
+      "loss": 0.0756,
+      "step": 24685
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7608966376089663e-05,
+      "loss": 0.0692,
+      "step": 24686
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.759651307596513e-05,
+      "loss": 0.1119,
+      "step": 24687
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7584059775840595e-05,
+      "loss": 0.1589,
+      "step": 24688
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7571606475716064e-05,
+      "loss": 0.101,
+      "step": 24689
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.755915317559153e-05,
+      "loss": 0.086,
+      "step": 24690
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7546699875466996e-05,
+      "loss": 0.1933,
+      "step": 24691
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7534246575342462e-05,
+      "loss": 0.0768,
+      "step": 24692
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.752179327521793e-05,
+      "loss": 0.2374,
+      "step": 24693
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.7509339975093397e-05,
+      "loss": 0.177,
+      "step": 24694
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7496886674968867e-05,
+      "loss": 0.1683,
+      "step": 24695
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7484433374844333e-05,
+      "loss": 0.089,
+      "step": 24696
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.74719800747198e-05,
+      "loss": 0.0737,
+      "step": 24697
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7459526774595264e-05,
+      "loss": 0.0495,
+      "step": 24698
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7447073474470734e-05,
+      "loss": 0.1137,
+      "step": 24699
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.74346201743462e-05,
+      "loss": 0.1326,
+      "step": 24700
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.742216687422167e-05,
+      "loss": 0.1708,
+      "step": 24701
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7409713574097135e-05,
+      "loss": 0.0793,
+      "step": 24702
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.73972602739726e-05,
+      "loss": 0.0991,
+      "step": 24703
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7384806973848067e-05,
+      "loss": 0.0697,
+      "step": 24704
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7372353673723536e-05,
+      "loss": 0.1141,
+      "step": 24705
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7359900373599002e-05,
+      "loss": 0.1263,
+      "step": 24706
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.734744707347447e-05,
+      "loss": 0.1421,
+      "step": 24707
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7334993773349937e-05,
+      "loss": 0.1667,
+      "step": 24708
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7322540473225403e-05,
+      "loss": 0.0919,
+      "step": 24709
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.731008717310087e-05,
+      "loss": 0.103,
+      "step": 24710
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.729763387297634e-05,
+      "loss": 0.0962,
+      "step": 24711
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7285180572851805e-05,
+      "loss": 0.0831,
+      "step": 24712
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.727272727272727e-05,
+      "loss": 0.2081,
+      "step": 24713
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7260273972602737e-05,
+      "loss": 0.1757,
+      "step": 24714
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7247820672478206e-05,
+      "loss": 0.1247,
+      "step": 24715
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7235367372353672e-05,
+      "loss": 0.1239,
+      "step": 24716
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.722291407222914e-05,
+      "loss": 0.0491,
+      "step": 24717
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7210460772104607e-05,
+      "loss": 0.0569,
+      "step": 24718
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.7198007471980073e-05,
+      "loss": 0.1221,
+      "step": 24719
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.718555417185554e-05,
+      "loss": 0.2067,
+      "step": 24720
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7173100871731005e-05,
+      "loss": 0.1391,
+      "step": 24721
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7160647571606474e-05,
+      "loss": 0.1095,
+      "step": 24722
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.714819427148194e-05,
+      "loss": 0.1331,
+      "step": 24723
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.713574097135741e-05,
+      "loss": 0.1475,
+      "step": 24724
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7123287671232875e-05,
+      "loss": 0.123,
+      "step": 24725
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.711083437110834e-05,
+      "loss": 0.1855,
+      "step": 24726
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7098381070983807e-05,
+      "loss": 0.1014,
+      "step": 24727
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7085927770859277e-05,
+      "loss": 0.1337,
+      "step": 24728
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7073474470734743e-05,
+      "loss": 0.0801,
+      "step": 24729
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7061021170610212e-05,
+      "loss": 0.1033,
+      "step": 24730
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7048567870485678e-05,
+      "loss": 0.1305,
+      "step": 24731
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.7036114570361144e-05,
+      "loss": 0.0912,
+      "step": 24732
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.702366127023661e-05,
+      "loss": 0.1342,
+      "step": 24733
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.701120797011208e-05,
+      "loss": 0.0942,
+      "step": 24734
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.6998754669987545e-05,
+      "loss": 0.0919,
+      "step": 24735
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.698630136986301e-05,
+      "loss": 0.0742,
+      "step": 24736
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.6973848069738477e-05,
+      "loss": 0.0659,
+      "step": 24737
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.6961394769613946e-05,
+      "loss": 0.1337,
+      "step": 24738
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.6948941469489412e-05,
+      "loss": 0.1046,
+      "step": 24739
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.693648816936488e-05,
+      "loss": 0.0951,
+      "step": 24740
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.6924034869240348e-05,
+      "loss": 0.0711,
+      "step": 24741
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.6911581569115813e-05,
+      "loss": 0.0621,
+      "step": 24742
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.689912826899128e-05,
+      "loss": 0.1032,
+      "step": 24743
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.688667496886675e-05,
+      "loss": 0.1985,
+      "step": 24744
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.6874221668742215e-05,
+      "loss": 0.1424,
+      "step": 24745
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.6861768368617684e-05,
+      "loss": 0.1009,
+      "step": 24746
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.684931506849315e-05,
+      "loss": 0.1214,
+      "step": 24747
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6836861768368616e-05,
+      "loss": 0.0677,
+      "step": 24748
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6824408468244082e-05,
+      "loss": 0.1806,
+      "step": 24749
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.681195516811955e-05,
+      "loss": 0.1497,
+      "step": 24750
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6799501867995017e-05,
+      "loss": 0.1484,
+      "step": 24751
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6787048567870486e-05,
+      "loss": 0.1038,
+      "step": 24752
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6774595267745952e-05,
+      "loss": 0.0907,
+      "step": 24753
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.676214196762142e-05,
+      "loss": 0.0763,
+      "step": 24754
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6749688667496884e-05,
+      "loss": 0.0557,
+      "step": 24755
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6737235367372354e-05,
+      "loss": 0.1182,
+      "step": 24756
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.672478206724782e-05,
+      "loss": 0.2087,
+      "step": 24757
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6712328767123286e-05,
+      "loss": 0.128,
+      "step": 24758
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.669987546699875e-05,
+      "loss": 0.0691,
+      "step": 24759
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.668742216687422e-05,
+      "loss": 0.0823,
+      "step": 24760
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6674968866749687e-05,
+      "loss": 0.1632,
+      "step": 24761
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6662515566625156e-05,
+      "loss": 0.1606,
+      "step": 24762
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6650062266500622e-05,
+      "loss": 0.1661,
+      "step": 24763
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6637608966376088e-05,
+      "loss": 0.1552,
+      "step": 24764
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6625155666251554e-05,
+      "loss": 0.15,
+      "step": 24765
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6612702366127023e-05,
+      "loss": 0.1464,
+      "step": 24766
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.660024906600249e-05,
+      "loss": 0.0958,
+      "step": 24767
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.658779576587796e-05,
+      "loss": 0.0582,
+      "step": 24768
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6575342465753425e-05,
+      "loss": 0.1267,
+      "step": 24769
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.656288916562889e-05,
+      "loss": 0.1739,
+      "step": 24770
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6550435865504356e-05,
+      "loss": 0.1203,
+      "step": 24771
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.6537982565379826e-05,
+      "loss": 0.1051,
+      "step": 24772
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.652552926525529e-05,
+      "loss": 0.1467,
+      "step": 24773
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6513075965130758e-05,
+      "loss": 0.0685,
+      "step": 24774
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6500622665006224e-05,
+      "loss": 0.18,
+      "step": 24775
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6488169364881693e-05,
+      "loss": 0.1375,
+      "step": 24776
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.647571606475716e-05,
+      "loss": 0.1438,
+      "step": 24777
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6463262764632625e-05,
+      "loss": 0.0967,
+      "step": 24778
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6450809464508094e-05,
+      "loss": 0.0665,
+      "step": 24779
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.643835616438356e-05,
+      "loss": 0.1022,
+      "step": 24780
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6425902864259026e-05,
+      "loss": 0.2071,
+      "step": 24781
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6413449564134492e-05,
+      "loss": 0.1396,
+      "step": 24782
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.640099626400996e-05,
+      "loss": 0.186,
+      "step": 24783
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6388542963885427e-05,
+      "loss": 0.1169,
+      "step": 24784
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6376089663760897e-05,
+      "loss": 0.1355,
+      "step": 24785
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6363636363636363e-05,
+      "loss": 0.0546,
+      "step": 24786
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.635118306351183e-05,
+      "loss": 0.154,
+      "step": 24787
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6338729763387294e-05,
+      "loss": 0.1729,
+      "step": 24788
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6326276463262764e-05,
+      "loss": 0.1666,
+      "step": 24789
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.631382316313823e-05,
+      "loss": 0.1035,
+      "step": 24790
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.63013698630137e-05,
+      "loss": 0.1667,
+      "step": 24791
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6288916562889165e-05,
+      "loss": 0.182,
+      "step": 24792
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.627646326276463e-05,
+      "loss": 0.0794,
+      "step": 24793
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6264009962640097e-05,
+      "loss": 0.1496,
+      "step": 24794
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6251556662515566e-05,
+      "loss": 0.1403,
+      "step": 24795
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6239103362391032e-05,
+      "loss": 0.2443,
+      "step": 24796
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6226650062266498e-05,
+      "loss": 0.1628,
+      "step": 24797
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.6214196762141964e-05,
+      "loss": 0.07,
+      "step": 24798
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6201743462017433e-05,
+      "loss": 0.1089,
+      "step": 24799
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.61892901618929e-05,
+      "loss": 0.2384,
+      "step": 24800
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.617683686176837e-05,
+      "loss": 0.1533,
+      "step": 24801
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6164383561643835e-05,
+      "loss": 0.1172,
+      "step": 24802
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.61519302615193e-05,
+      "loss": 0.1054,
+      "step": 24803
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6139476961394766e-05,
+      "loss": 0.1125,
+      "step": 24804
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6127023661270236e-05,
+      "loss": 0.0935,
+      "step": 24805
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6114570361145702e-05,
+      "loss": 0.1534,
+      "step": 24806
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.610211706102117e-05,
+      "loss": 0.1352,
+      "step": 24807
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6089663760896637e-05,
+      "loss": 0.1153,
+      "step": 24808
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6077210460772103e-05,
+      "loss": 0.1472,
+      "step": 24809
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.606475716064757e-05,
+      "loss": 0.1505,
+      "step": 24810
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6052303860523038e-05,
+      "loss": 0.1197,
+      "step": 24811
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6039850560398504e-05,
+      "loss": 0.0973,
+      "step": 24812
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6027397260273974e-05,
+      "loss": 0.1484,
+      "step": 24813
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.601494396014944e-05,
+      "loss": 0.1406,
+      "step": 24814
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.6002490660024905e-05,
+      "loss": 0.2749,
+      "step": 24815
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.599003735990037e-05,
+      "loss": 0.1269,
+      "step": 24816
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.597758405977584e-05,
+      "loss": 0.1495,
+      "step": 24817
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.5965130759651307e-05,
+      "loss": 0.0736,
+      "step": 24818
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.5952677459526773e-05,
+      "loss": 0.1393,
+      "step": 24819
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.594022415940224e-05,
+      "loss": 0.0783,
+      "step": 24820
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.5927770859277708e-05,
+      "loss": 0.1326,
+      "step": 24821
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.5915317559153174e-05,
+      "loss": 0.1584,
+      "step": 24822
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.5902864259028643e-05,
+      "loss": 0.0693,
+      "step": 24823
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.589041095890411e-05,
+      "loss": 0.1327,
+      "step": 24824
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5877957658779575e-05,
+      "loss": 0.2929,
+      "step": 24825
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.586550435865504e-05,
+      "loss": 0.2025,
+      "step": 24826
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5853051058530507e-05,
+      "loss": 0.0884,
+      "step": 24827
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5840597758405976e-05,
+      "loss": 0.0871,
+      "step": 24828
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5828144458281442e-05,
+      "loss": 0.1217,
+      "step": 24829
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.581569115815691e-05,
+      "loss": 0.0664,
+      "step": 24830
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5803237858032377e-05,
+      "loss": 0.1222,
+      "step": 24831
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5790784557907843e-05,
+      "loss": 0.1162,
+      "step": 24832
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.577833125778331e-05,
+      "loss": 0.1223,
+      "step": 24833
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.576587795765878e-05,
+      "loss": 0.1076,
+      "step": 24834
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5753424657534245e-05,
+      "loss": 0.1204,
+      "step": 24835
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5740971357409714e-05,
+      "loss": 0.0806,
+      "step": 24836
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.572851805728518e-05,
+      "loss": 0.1129,
+      "step": 24837
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5716064757160646e-05,
+      "loss": 0.2036,
+      "step": 24838
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5703611457036112e-05,
+      "loss": 0.1401,
+      "step": 24839
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.569115815691158e-05,
+      "loss": 0.1042,
+      "step": 24840
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5678704856787047e-05,
+      "loss": 0.0951,
+      "step": 24841
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5666251556662513e-05,
+      "loss": 0.1126,
+      "step": 24842
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.565379825653798e-05,
+      "loss": 0.1117,
+      "step": 24843
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.564134495641345e-05,
+      "loss": 0.121,
+      "step": 24844
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5628891656288914e-05,
+      "loss": 0.1819,
+      "step": 24845
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5616438356164384e-05,
+      "loss": 0.0832,
+      "step": 24846
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.560398505603985e-05,
+      "loss": 0.07,
+      "step": 24847
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.5591531755915315e-05,
+      "loss": 0.1324,
+      "step": 24848
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.557907845579078e-05,
+      "loss": 0.1268,
+      "step": 24849
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.556662515566625e-05,
+      "loss": 0.1755,
+      "step": 24850
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5554171855541717e-05,
+      "loss": 0.1443,
+      "step": 24851
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5541718555417186e-05,
+      "loss": 0.1424,
+      "step": 24852
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5529265255292652e-05,
+      "loss": 0.1151,
+      "step": 24853
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5516811955168118e-05,
+      "loss": 0.0878,
+      "step": 24854
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5504358655043584e-05,
+      "loss": 0.1287,
+      "step": 24855
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5491905354919053e-05,
+      "loss": 0.1339,
+      "step": 24856
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.547945205479452e-05,
+      "loss": 0.1278,
+      "step": 24857
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.546699875466999e-05,
+      "loss": 0.1082,
+      "step": 24858
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5454545454545454e-05,
+      "loss": 0.0678,
+      "step": 24859
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.544209215442092e-05,
+      "loss": 0.1415,
+      "step": 24860
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5429638854296386e-05,
+      "loss": 0.0783,
+      "step": 24861
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5417185554171856e-05,
+      "loss": 0.1355,
+      "step": 24862
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.540473225404732e-05,
+      "loss": 0.1591,
+      "step": 24863
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5392278953922788e-05,
+      "loss": 0.1057,
+      "step": 24864
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5379825653798253e-05,
+      "loss": 0.1571,
+      "step": 24865
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5367372353673723e-05,
+      "loss": 0.1275,
+      "step": 24866
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.535491905354919e-05,
+      "loss": 0.1203,
+      "step": 24867
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5342465753424658e-05,
+      "loss": 0.1833,
+      "step": 24868
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5330012453300124e-05,
+      "loss": 0.1681,
+      "step": 24869
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.531755915317559e-05,
+      "loss": 0.2051,
+      "step": 24870
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5305105853051056e-05,
+      "loss": 0.0907,
+      "step": 24871
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5292652552926525e-05,
+      "loss": 0.0966,
+      "step": 24872
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.528019925280199e-05,
+      "loss": 0.0602,
+      "step": 24873
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.526774595267746e-05,
+      "loss": 0.1117,
+      "step": 24874
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5255292652552925e-05,
+      "loss": 0.069,
+      "step": 24875
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.5242839352428392e-05,
+      "loss": 0.116,
+      "step": 24876
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5230386052303858e-05,
+      "loss": 0.104,
+      "step": 24877
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5217932752179328e-05,
+      "loss": 0.165,
+      "step": 24878
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5205479452054794e-05,
+      "loss": 0.1011,
+      "step": 24879
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.519302615193026e-05,
+      "loss": 0.1362,
+      "step": 24880
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5180572851805727e-05,
+      "loss": 0.1209,
+      "step": 24881
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5168119551681193e-05,
+      "loss": 0.1163,
+      "step": 24882
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.515566625155666e-05,
+      "loss": 0.1631,
+      "step": 24883
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5143212951432127e-05,
+      "loss": 0.1768,
+      "step": 24884
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5130759651307596e-05,
+      "loss": 0.086,
+      "step": 24885
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5118306351183062e-05,
+      "loss": 0.1592,
+      "step": 24886
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.510585305105853e-05,
+      "loss": 0.0987,
+      "step": 24887
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5093399750933996e-05,
+      "loss": 0.1418,
+      "step": 24888
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5080946450809463e-05,
+      "loss": 0.1516,
+      "step": 24889
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.506849315068493e-05,
+      "loss": 0.0725,
+      "step": 24890
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5056039850560397e-05,
+      "loss": 0.1295,
+      "step": 24891
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5043586550435863e-05,
+      "loss": 0.1092,
+      "step": 24892
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5031133250311332e-05,
+      "loss": 0.0672,
+      "step": 24893
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5018679950186798e-05,
+      "loss": 0.1628,
+      "step": 24894
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.5006226650062266e-05,
+      "loss": 0.1338,
+      "step": 24895
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.4993773349937732e-05,
+      "loss": 0.1277,
+      "step": 24896
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.49813200498132e-05,
+      "loss": 0.1323,
+      "step": 24897
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.4968866749688665e-05,
+      "loss": 0.0665,
+      "step": 24898
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.4956413449564133e-05,
+      "loss": 0.161,
+      "step": 24899
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.4943960149439599e-05,
+      "loss": 0.1423,
+      "step": 24900
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.4931506849315068e-05,
+      "loss": 0.1087,
+      "step": 24901
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.4919053549190534e-05,
+      "loss": 0.1344,
+      "step": 24902
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4906600249066002e-05,
+      "loss": 0.1145,
+      "step": 24903
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4894146948941468e-05,
+      "loss": 0.098,
+      "step": 24904
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4881693648816935e-05,
+      "loss": 0.0844,
+      "step": 24905
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4869240348692401e-05,
+      "loss": 0.1623,
+      "step": 24906
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4856787048567869e-05,
+      "loss": 0.1664,
+      "step": 24907
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4844333748443335e-05,
+      "loss": 0.2183,
+      "step": 24908
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4831880448318804e-05,
+      "loss": 0.068,
+      "step": 24909
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.481942714819427e-05,
+      "loss": 0.0725,
+      "step": 24910
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4806973848069738e-05,
+      "loss": 0.0683,
+      "step": 24911
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4794520547945204e-05,
+      "loss": 0.1209,
+      "step": 24912
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4782067247820671e-05,
+      "loss": 0.1969,
+      "step": 24913
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4769613947696137e-05,
+      "loss": 0.1127,
+      "step": 24914
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4757160647571607e-05,
+      "loss": 0.0919,
+      "step": 24915
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4744707347447073e-05,
+      "loss": 0.0967,
+      "step": 24916
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.473225404732254e-05,
+      "loss": 0.0979,
+      "step": 24917
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4719800747198006e-05,
+      "loss": 0.1716,
+      "step": 24918
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4707347447073474e-05,
+      "loss": 0.1138,
+      "step": 24919
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.469489414694894e-05,
+      "loss": 0.1347,
+      "step": 24920
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4682440846824407e-05,
+      "loss": 0.1794,
+      "step": 24921
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4669987546699873e-05,
+      "loss": 0.1251,
+      "step": 24922
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4657534246575343e-05,
+      "loss": 0.0878,
+      "step": 24923
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4645080946450809e-05,
+      "loss": 0.1139,
+      "step": 24924
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4632627646326276e-05,
+      "loss": 0.1128,
+      "step": 24925
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4620174346201742e-05,
+      "loss": 0.1859,
+      "step": 24926
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.460772104607721e-05,
+      "loss": 0.1075,
+      "step": 24927
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.4595267745952676e-05,
+      "loss": 0.1484,
+      "step": 24928
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4582814445828143e-05,
+      "loss": 0.0905,
+      "step": 24929
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.457036114570361e-05,
+      "loss": 0.1715,
+      "step": 24930
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4557907845579079e-05,
+      "loss": 0.1132,
+      "step": 24931
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4545454545454545e-05,
+      "loss": 0.1711,
+      "step": 24932
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.453300124533001e-05,
+      "loss": 0.098,
+      "step": 24933
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4520547945205478e-05,
+      "loss": 0.1319,
+      "step": 24934
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4508094645080944e-05,
+      "loss": 0.0679,
+      "step": 24935
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4495641344956412e-05,
+      "loss": 0.0834,
+      "step": 24936
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4483188044831878e-05,
+      "loss": 0.1679,
+      "step": 24937
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4470734744707347e-05,
+      "loss": 0.1798,
+      "step": 24938
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4458281444582813e-05,
+      "loss": 0.2362,
+      "step": 24939
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.444582814445828e-05,
+      "loss": 0.1402,
+      "step": 24940
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4433374844333747e-05,
+      "loss": 0.0591,
+      "step": 24941
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4420921544209214e-05,
+      "loss": 0.1401,
+      "step": 24942
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.440846824408468e-05,
+      "loss": 0.0831,
+      "step": 24943
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4396014943960148e-05,
+      "loss": 0.1572,
+      "step": 24944
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4383561643835614e-05,
+      "loss": 0.2011,
+      "step": 24945
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4371108343711083e-05,
+      "loss": 0.1338,
+      "step": 24946
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4358655043586549e-05,
+      "loss": 0.0988,
+      "step": 24947
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4346201743462017e-05,
+      "loss": 0.0856,
+      "step": 24948
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4333748443337483e-05,
+      "loss": 0.1398,
+      "step": 24949
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.432129514321295e-05,
+      "loss": 0.0924,
+      "step": 24950
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4308841843088416e-05,
+      "loss": 0.1756,
+      "step": 24951
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4296388542963884e-05,
+      "loss": 0.1385,
+      "step": 24952
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.428393524283935e-05,
+      "loss": 0.0828,
+      "step": 24953
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4271481942714819e-05,
+      "loss": 0.0526,
+      "step": 24954
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.4259028642590285e-05,
+      "loss": 0.1056,
+      "step": 24955
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4246575342465753e-05,
+      "loss": 0.2068,
+      "step": 24956
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4234122042341219e-05,
+      "loss": 0.1535,
+      "step": 24957
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4221668742216686e-05,
+      "loss": 0.1079,
+      "step": 24958
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4209215442092152e-05,
+      "loss": 0.1451,
+      "step": 24959
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.419676214196762e-05,
+      "loss": 0.0791,
+      "step": 24960
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4184308841843086e-05,
+      "loss": 0.1022,
+      "step": 24961
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4171855541718555e-05,
+      "loss": 0.1408,
+      "step": 24962
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4159402241594021e-05,
+      "loss": 0.1975,
+      "step": 24963
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4146948941469489e-05,
+      "loss": 0.1287,
+      "step": 24964
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4134495641344955e-05,
+      "loss": 0.0841,
+      "step": 24965
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4122042341220422e-05,
+      "loss": 0.0723,
+      "step": 24966
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4109589041095888e-05,
+      "loss": 0.1987,
+      "step": 24967
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4097135740971358e-05,
+      "loss": 0.1362,
+      "step": 24968
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4084682440846824e-05,
+      "loss": 0.1003,
+      "step": 24969
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4072229140722291e-05,
+      "loss": 0.1124,
+      "step": 24970
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4059775840597757e-05,
+      "loss": 0.1536,
+      "step": 24971
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4047322540473225e-05,
+      "loss": 0.1062,
+      "step": 24972
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.403486924034869e-05,
+      "loss": 0.0788,
+      "step": 24973
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4022415940224158e-05,
+      "loss": 0.1564,
+      "step": 24974
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.4009962640099624e-05,
+      "loss": 0.1276,
+      "step": 24975
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3997509339975094e-05,
+      "loss": 0.1819,
+      "step": 24976
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.398505603985056e-05,
+      "loss": 0.1962,
+      "step": 24977
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3972602739726027e-05,
+      "loss": 0.0685,
+      "step": 24978
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3960149439601493e-05,
+      "loss": 0.1082,
+      "step": 24979
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.394769613947696e-05,
+      "loss": 0.0453,
+      "step": 24980
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.3935242839352427e-05,
+      "loss": 0.1599,
+      "step": 24981
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3922789539227894e-05,
+      "loss": 0.1195,
+      "step": 24982
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.391033623910336e-05,
+      "loss": 0.1193,
+      "step": 24983
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.389788293897883e-05,
+      "loss": 0.2215,
+      "step": 24984
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3885429638854296e-05,
+      "loss": 0.1355,
+      "step": 24985
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3872976338729762e-05,
+      "loss": 0.0573,
+      "step": 24986
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.386052303860523e-05,
+      "loss": 0.0898,
+      "step": 24987
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3848069738480695e-05,
+      "loss": 0.2811,
+      "step": 24988
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3835616438356163e-05,
+      "loss": 0.197,
+      "step": 24989
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3823163138231629e-05,
+      "loss": 0.2316,
+      "step": 24990
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3810709838107098e-05,
+      "loss": 0.0991,
+      "step": 24991
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3798256537982564e-05,
+      "loss": 0.0638,
+      "step": 24992
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3785803237858032e-05,
+      "loss": 0.1107,
+      "step": 24993
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3773349937733498e-05,
+      "loss": 0.1585,
+      "step": 24994
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3760896637608965e-05,
+      "loss": 0.2108,
+      "step": 24995
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3748443337484431e-05,
+      "loss": 0.1763,
+      "step": 24996
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3735990037359899e-05,
+      "loss": 0.1088,
+      "step": 24997
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3723536737235365e-05,
+      "loss": 0.076,
+      "step": 24998
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3711083437110834e-05,
+      "loss": 0.1081,
+      "step": 24999
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.36986301369863e-05,
+      "loss": 0.1141,
+      "step": 25000
+    },
+    {
+      "epoch": 9.58,
+      "eval_cer": 0.1002009481631777,
+      "eval_loss": 0.4263627231121063,
+      "eval_runtime": 47.8736,
+      "eval_samples_per_second": 36.617,
+      "eval_steps_per_second": 4.595,
+      "eval_wer": 0.15793999185409785,
+      "step": 25000
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3686176836861768e-05,
+      "loss": 0.176,
+      "step": 25001
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3673723536737234e-05,
+      "loss": 0.0951,
+      "step": 25002
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3661270236612701e-05,
+      "loss": 0.1192,
+      "step": 25003
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3648816936488167e-05,
+      "loss": 0.0679,
+      "step": 25004
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.3636363636363635e-05,
+      "loss": 0.0572,
+      "step": 25005
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.36239103362391e-05,
+      "loss": 0.2664,
+      "step": 25006
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.361145703611457e-05,
+      "loss": 0.1473,
+      "step": 25007
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3599003735990036e-05,
+      "loss": 0.1773,
+      "step": 25008
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3586550435865504e-05,
+      "loss": 0.1959,
+      "step": 25009
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.357409713574097e-05,
+      "loss": 0.0756,
+      "step": 25010
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3561643835616437e-05,
+      "loss": 0.092,
+      "step": 25011
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3549190535491903e-05,
+      "loss": 0.0979,
+      "step": 25012
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3536737235367371e-05,
+      "loss": 0.1675,
+      "step": 25013
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3524283935242837e-05,
+      "loss": 0.1263,
+      "step": 25014
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3511830635118306e-05,
+      "loss": 0.0871,
+      "step": 25015
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3499377334993772e-05,
+      "loss": 0.1704,
+      "step": 25016
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.348692403486924e-05,
+      "loss": 0.1037,
+      "step": 25017
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3474470734744706e-05,
+      "loss": 0.257,
+      "step": 25018
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3462017434620173e-05,
+      "loss": 0.1448,
+      "step": 25019
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.344956413449564e-05,
+      "loss": 0.1206,
+      "step": 25020
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3437110834371109e-05,
+      "loss": 0.0741,
+      "step": 25021
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3424657534246575e-05,
+      "loss": 0.0875,
+      "step": 25022
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3412204234122042e-05,
+      "loss": 0.0844,
+      "step": 25023
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3399750933997508e-05,
+      "loss": 0.083,
+      "step": 25024
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3387297633872976e-05,
+      "loss": 0.1073,
+      "step": 25025
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3374844333748442e-05,
+      "loss": 0.1197,
+      "step": 25026
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.336239103362391e-05,
+      "loss": 0.0763,
+      "step": 25027
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3349937733499375e-05,
+      "loss": 0.2256,
+      "step": 25028
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3337484433374845e-05,
+      "loss": 0.0815,
+      "step": 25029
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.332503113325031e-05,
+      "loss": 0.1772,
+      "step": 25030
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3312577833125778e-05,
+      "loss": 0.1469,
+      "step": 25031
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3300124533001244e-05,
+      "loss": 0.2309,
+      "step": 25032
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.3287671232876712e-05,
+      "loss": 0.0986,
+      "step": 25033
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3275217932752178e-05,
+      "loss": 0.0926,
+      "step": 25034
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3262764632627645e-05,
+      "loss": 0.0933,
+      "step": 25035
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3250311332503111e-05,
+      "loss": 0.0621,
+      "step": 25036
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.323785803237858e-05,
+      "loss": 0.0887,
+      "step": 25037
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3225404732254047e-05,
+      "loss": 0.1377,
+      "step": 25038
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3212951432129513e-05,
+      "loss": 0.1101,
+      "step": 25039
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.320049813200498e-05,
+      "loss": 0.1054,
+      "step": 25040
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3188044831880446e-05,
+      "loss": 0.1604,
+      "step": 25041
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3175591531755914e-05,
+      "loss": 0.0662,
+      "step": 25042
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.316313823163138e-05,
+      "loss": 0.0807,
+      "step": 25043
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3150684931506849e-05,
+      "loss": 0.1252,
+      "step": 25044
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3138231631382315e-05,
+      "loss": 0.1316,
+      "step": 25045
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3125778331257783e-05,
+      "loss": 0.1599,
+      "step": 25046
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3113325031133249e-05,
+      "loss": 0.1602,
+      "step": 25047
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3100871731008716e-05,
+      "loss": 0.1432,
+      "step": 25048
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3088418430884182e-05,
+      "loss": 0.0666,
+      "step": 25049
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.307596513075965e-05,
+      "loss": 0.1582,
+      "step": 25050
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3063511830635116e-05,
+      "loss": 0.1336,
+      "step": 25051
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3051058530510585e-05,
+      "loss": 0.0958,
+      "step": 25052
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3038605230386051e-05,
+      "loss": 0.1219,
+      "step": 25053
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3026151930261519e-05,
+      "loss": 0.156,
+      "step": 25054
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3013698630136985e-05,
+      "loss": 0.0736,
+      "step": 25055
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.3001245330012452e-05,
+      "loss": 0.1704,
+      "step": 25056
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.2988792029887918e-05,
+      "loss": 0.1702,
+      "step": 25057
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.2976338729763386e-05,
+      "loss": 0.1599,
+      "step": 25058
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.2963885429638852e-05,
+      "loss": 0.1602,
+      "step": 25059
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2951432129514321e-05,
+      "loss": 0.1087,
+      "step": 25060
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2938978829389787e-05,
+      "loss": 0.0698,
+      "step": 25061
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2926525529265255e-05,
+      "loss": 0.0683,
+      "step": 25062
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.291407222914072e-05,
+      "loss": 0.2744,
+      "step": 25063
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2901618929016188e-05,
+      "loss": 0.1209,
+      "step": 25064
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2889165628891654e-05,
+      "loss": 0.0794,
+      "step": 25065
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2876712328767122e-05,
+      "loss": 0.0677,
+      "step": 25066
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2864259028642588e-05,
+      "loss": 0.1094,
+      "step": 25067
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2851805728518057e-05,
+      "loss": 0.107,
+      "step": 25068
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2839352428393523e-05,
+      "loss": 0.1421,
+      "step": 25069
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.282689912826899e-05,
+      "loss": 0.105,
+      "step": 25070
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2814445828144457e-05,
+      "loss": 0.0792,
+      "step": 25071
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2801992528019924e-05,
+      "loss": 0.0686,
+      "step": 25072
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.278953922789539e-05,
+      "loss": 0.1156,
+      "step": 25073
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.277708592777086e-05,
+      "loss": 0.1187,
+      "step": 25074
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2764632627646326e-05,
+      "loss": 0.1286,
+      "step": 25075
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2752179327521793e-05,
+      "loss": 0.1487,
+      "step": 25076
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2739726027397259e-05,
+      "loss": 0.1357,
+      "step": 25077
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2727272727272727e-05,
+      "loss": 0.1753,
+      "step": 25078
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2714819427148193e-05,
+      "loss": 0.1314,
+      "step": 25079
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.270236612702366e-05,
+      "loss": 0.0634,
+      "step": 25080
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2689912826899126e-05,
+      "loss": 0.1121,
+      "step": 25081
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2677459526774596e-05,
+      "loss": 0.1911,
+      "step": 25082
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2665006226650062e-05,
+      "loss": 0.1248,
+      "step": 25083
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.265255292652553e-05,
+      "loss": 0.093,
+      "step": 25084
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.2640099626400995e-05,
+      "loss": 0.0739,
+      "step": 25085
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2627646326276463e-05,
+      "loss": 0.086,
+      "step": 25086
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2615193026151929e-05,
+      "loss": 0.2599,
+      "step": 25087
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2602739726027396e-05,
+      "loss": 0.177,
+      "step": 25088
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2590286425902862e-05,
+      "loss": 0.1019,
+      "step": 25089
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2577833125778332e-05,
+      "loss": 0.1175,
+      "step": 25090
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2565379825653798e-05,
+      "loss": 0.0677,
+      "step": 25091
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2552926525529264e-05,
+      "loss": 0.0707,
+      "step": 25092
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2540473225404731e-05,
+      "loss": 0.1574,
+      "step": 25093
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2528019925280197e-05,
+      "loss": 0.2002,
+      "step": 25094
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2515566625155665e-05,
+      "loss": 0.1059,
+      "step": 25095
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.250311332503113e-05,
+      "loss": 0.0759,
+      "step": 25096
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.24906600249066e-05,
+      "loss": 0.241,
+      "step": 25097
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2478206724782066e-05,
+      "loss": 0.1002,
+      "step": 25098
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2465753424657534e-05,
+      "loss": 0.0458,
+      "step": 25099
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2453300124533e-05,
+      "loss": 0.1477,
+      "step": 25100
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2440846824408467e-05,
+      "loss": 0.1236,
+      "step": 25101
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2428393524283933e-05,
+      "loss": 0.0917,
+      "step": 25102
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.24159402241594e-05,
+      "loss": 0.0918,
+      "step": 25103
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2403486924034867e-05,
+      "loss": 0.1972,
+      "step": 25104
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2391033623910336e-05,
+      "loss": 0.1262,
+      "step": 25105
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2378580323785802e-05,
+      "loss": 0.181,
+      "step": 25106
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.236612702366127e-05,
+      "loss": 0.1772,
+      "step": 25107
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2353673723536736e-05,
+      "loss": 0.0916,
+      "step": 25108
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2341220423412203e-05,
+      "loss": 0.1387,
+      "step": 25109
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.232876712328767e-05,
+      "loss": 0.1931,
+      "step": 25110
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.2316313823163137e-05,
+      "loss": 0.1162,
+      "step": 25111
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2303860523038603e-05,
+      "loss": 0.1401,
+      "step": 25112
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2291407222914072e-05,
+      "loss": 0.2364,
+      "step": 25113
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2278953922789538e-05,
+      "loss": 0.168,
+      "step": 25114
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2266500622665006e-05,
+      "loss": 0.1183,
+      "step": 25115
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2254047322540472e-05,
+      "loss": 0.1342,
+      "step": 25116
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.224159402241594e-05,
+      "loss": 0.1606,
+      "step": 25117
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2229140722291405e-05,
+      "loss": 0.1018,
+      "step": 25118
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2216687422166873e-05,
+      "loss": 0.1851,
+      "step": 25119
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.220423412204234e-05,
+      "loss": 0.1578,
+      "step": 25120
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2191780821917808e-05,
+      "loss": 0.1158,
+      "step": 25121
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2179327521793274e-05,
+      "loss": 0.0885,
+      "step": 25122
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2166874221668742e-05,
+      "loss": 0.1233,
+      "step": 25123
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2154420921544208e-05,
+      "loss": 0.0543,
+      "step": 25124
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2141967621419675e-05,
+      "loss": 0.1367,
+      "step": 25125
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2129514321295141e-05,
+      "loss": 0.1415,
+      "step": 25126
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.211706102117061e-05,
+      "loss": 0.1019,
+      "step": 25127
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2104607721046077e-05,
+      "loss": 0.1894,
+      "step": 25128
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2092154420921544e-05,
+      "loss": 0.0933,
+      "step": 25129
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.207970112079701e-05,
+      "loss": 0.0491,
+      "step": 25130
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2067247820672478e-05,
+      "loss": 0.2397,
+      "step": 25131
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2054794520547944e-05,
+      "loss": 0.1754,
+      "step": 25132
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2042341220423411e-05,
+      "loss": 0.1615,
+      "step": 25133
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2029887920298877e-05,
+      "loss": 0.138,
+      "step": 25134
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2017434620174347e-05,
+      "loss": 0.1692,
+      "step": 25135
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.2004981320049813e-05,
+      "loss": 0.0739,
+      "step": 25136
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.199252801992528e-05,
+      "loss": 0.1375,
+      "step": 25137
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1980074719800746e-05,
+      "loss": 0.1856,
+      "step": 25138
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1967621419676214e-05,
+      "loss": 0.1626,
+      "step": 25139
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.195516811955168e-05,
+      "loss": 0.0998,
+      "step": 25140
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1942714819427147e-05,
+      "loss": 0.0668,
+      "step": 25141
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1930261519302613e-05,
+      "loss": 0.0562,
+      "step": 25142
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1917808219178083e-05,
+      "loss": 0.1069,
+      "step": 25143
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1905354919053549e-05,
+      "loss": 0.2444,
+      "step": 25144
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1892901618929015e-05,
+      "loss": 0.1217,
+      "step": 25145
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1880448318804482e-05,
+      "loss": 0.135,
+      "step": 25146
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1867995018679948e-05,
+      "loss": 0.1087,
+      "step": 25147
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1855541718555416e-05,
+      "loss": 0.0729,
+      "step": 25148
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1843088418430882e-05,
+      "loss": 0.1324,
+      "step": 25149
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1830635118306351e-05,
+      "loss": 0.1751,
+      "step": 25150
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1818181818181817e-05,
+      "loss": 0.1762,
+      "step": 25151
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1805728518057285e-05,
+      "loss": 0.1797,
+      "step": 25152
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.179327521793275e-05,
+      "loss": 0.088,
+      "step": 25153
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1780821917808218e-05,
+      "loss": 0.1237,
+      "step": 25154
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1768368617683684e-05,
+      "loss": 0.0434,
+      "step": 25155
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1755915317559152e-05,
+      "loss": 0.0772,
+      "step": 25156
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1743462017434618e-05,
+      "loss": 0.1155,
+      "step": 25157
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1731008717310087e-05,
+      "loss": 0.1104,
+      "step": 25158
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1718555417185553e-05,
+      "loss": 0.0822,
+      "step": 25159
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.170610211706102e-05,
+      "loss": 0.0826,
+      "step": 25160
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1693648816936487e-05,
+      "loss": 0.1098,
+      "step": 25161
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.1681195516811954e-05,
+      "loss": 0.1381,
+      "step": 25162
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.166874221668742e-05,
+      "loss": 0.1594,
+      "step": 25163
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1656288916562888e-05,
+      "loss": 0.0986,
+      "step": 25164
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1643835616438354e-05,
+      "loss": 0.1834,
+      "step": 25165
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1631382316313823e-05,
+      "loss": 0.1527,
+      "step": 25166
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1618929016189289e-05,
+      "loss": 0.1068,
+      "step": 25167
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1606475716064757e-05,
+      "loss": 0.1277,
+      "step": 25168
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1594022415940223e-05,
+      "loss": 0.2124,
+      "step": 25169
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.158156911581569e-05,
+      "loss": 0.1117,
+      "step": 25170
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1569115815691156e-05,
+      "loss": 0.0758,
+      "step": 25171
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1556662515566624e-05,
+      "loss": 0.0644,
+      "step": 25172
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1544209215442091e-05,
+      "loss": 0.1023,
+      "step": 25173
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1531755915317559e-05,
+      "loss": 0.1209,
+      "step": 25174
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1519302615193025e-05,
+      "loss": 0.2549,
+      "step": 25175
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1506849315068493e-05,
+      "loss": 0.1439,
+      "step": 25176
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1494396014943959e-05,
+      "loss": 0.0889,
+      "step": 25177
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1481942714819426e-05,
+      "loss": 0.0861,
+      "step": 25178
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1469489414694892e-05,
+      "loss": 0.158,
+      "step": 25179
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1457036114570362e-05,
+      "loss": 0.1245,
+      "step": 25180
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1444582814445828e-05,
+      "loss": 0.173,
+      "step": 25181
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1432129514321295e-05,
+      "loss": 0.0848,
+      "step": 25182
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1419676214196761e-05,
+      "loss": 0.0638,
+      "step": 25183
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1407222914072229e-05,
+      "loss": 0.0748,
+      "step": 25184
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1394769613947695e-05,
+      "loss": 0.1032,
+      "step": 25185
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1382316313823162e-05,
+      "loss": 0.1397,
+      "step": 25186
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1369863013698628e-05,
+      "loss": 0.1586,
+      "step": 25187
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1357409713574098e-05,
+      "loss": 0.1237,
+      "step": 25188
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.1344956413449564e-05,
+      "loss": 0.1566,
+      "step": 25189
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1332503113325031e-05,
+      "loss": 0.154,
+      "step": 25190
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1320049813200497e-05,
+      "loss": 0.0905,
+      "step": 25191
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1307596513075965e-05,
+      "loss": 0.0628,
+      "step": 25192
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.129514321295143e-05,
+      "loss": 0.181,
+      "step": 25193
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1282689912826898e-05,
+      "loss": 0.2266,
+      "step": 25194
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1270236612702364e-05,
+      "loss": 0.1756,
+      "step": 25195
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1257783312577834e-05,
+      "loss": 0.1155,
+      "step": 25196
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.12453300124533e-05,
+      "loss": 0.1395,
+      "step": 25197
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1232876712328766e-05,
+      "loss": 0.1252,
+      "step": 25198
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1220423412204233e-05,
+      "loss": 0.0861,
+      "step": 25199
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1207970112079699e-05,
+      "loss": 0.2,
+      "step": 25200
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1195516811955167e-05,
+      "loss": 0.1034,
+      "step": 25201
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1183063511830633e-05,
+      "loss": 0.1337,
+      "step": 25202
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1170610211706102e-05,
+      "loss": 0.1034,
+      "step": 25203
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1158156911581568e-05,
+      "loss": 0.1154,
+      "step": 25204
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1145703611457036e-05,
+      "loss": 0.1741,
+      "step": 25205
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1133250311332502e-05,
+      "loss": 0.1821,
+      "step": 25206
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.112079701120797e-05,
+      "loss": 0.2462,
+      "step": 25207
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1108343711083435e-05,
+      "loss": 0.1345,
+      "step": 25208
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1095890410958903e-05,
+      "loss": 0.0828,
+      "step": 25209
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1083437110834369e-05,
+      "loss": 0.0595,
+      "step": 25210
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1070983810709838e-05,
+      "loss": 0.0905,
+      "step": 25211
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1058530510585304e-05,
+      "loss": 0.0838,
+      "step": 25212
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1046077210460772e-05,
+      "loss": 0.191,
+      "step": 25213
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1033623910336238e-05,
+      "loss": 0.184,
+      "step": 25214
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.1021170610211705e-05,
+      "loss": 0.0999,
+      "step": 25215
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.1008717310087171e-05,
+      "loss": 0.0814,
+      "step": 25216
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0996264009962639e-05,
+      "loss": 0.1291,
+      "step": 25217
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0983810709838105e-05,
+      "loss": 0.1227,
+      "step": 25218
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0971357409713574e-05,
+      "loss": 0.1863,
+      "step": 25219
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.095890410958904e-05,
+      "loss": 0.1571,
+      "step": 25220
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0946450809464508e-05,
+      "loss": 0.0929,
+      "step": 25221
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0933997509339974e-05,
+      "loss": 0.1052,
+      "step": 25222
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0921544209215441e-05,
+      "loss": 0.0696,
+      "step": 25223
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0909090909090907e-05,
+      "loss": 0.1199,
+      "step": 25224
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0896637608966375e-05,
+      "loss": 0.1497,
+      "step": 25225
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0884184308841842e-05,
+      "loss": 0.1107,
+      "step": 25226
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.087173100871731e-05,
+      "loss": 0.1465,
+      "step": 25227
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0859277708592776e-05,
+      "loss": 0.1328,
+      "step": 25228
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0846824408468244e-05,
+      "loss": 0.0997,
+      "step": 25229
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.083437110834371e-05,
+      "loss": 0.0799,
+      "step": 25230
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0821917808219177e-05,
+      "loss": 0.2788,
+      "step": 25231
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0809464508094643e-05,
+      "loss": 0.212,
+      "step": 25232
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0797011207970113e-05,
+      "loss": 0.1327,
+      "step": 25233
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0784557907845579e-05,
+      "loss": 0.0749,
+      "step": 25234
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0772104607721046e-05,
+      "loss": 0.0518,
+      "step": 25235
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0759651307596512e-05,
+      "loss": 0.0906,
+      "step": 25236
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.074719800747198e-05,
+      "loss": 0.1458,
+      "step": 25237
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0734744707347446e-05,
+      "loss": 0.1748,
+      "step": 25238
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0722291407222913e-05,
+      "loss": 0.1121,
+      "step": 25239
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.070983810709838e-05,
+      "loss": 0.1912,
+      "step": 25240
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0697384806973849e-05,
+      "loss": 0.1049,
+      "step": 25241
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0684931506849315e-05,
+      "loss": 0.1412,
+      "step": 25242
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0672478206724782e-05,
+      "loss": 0.115,
+      "step": 25243
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0660024906600248e-05,
+      "loss": 0.1342,
+      "step": 25244
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0647571606475716e-05,
+      "loss": 0.1093,
+      "step": 25245
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0635118306351182e-05,
+      "loss": 0.1041,
+      "step": 25246
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.062266500622665e-05,
+      "loss": 0.121,
+      "step": 25247
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0610211706102115e-05,
+      "loss": 0.0705,
+      "step": 25248
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0597758405977585e-05,
+      "loss": 0.1057,
+      "step": 25249
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.058530510585305e-05,
+      "loss": 0.2289,
+      "step": 25250
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0572851805728518e-05,
+      "loss": 0.1813,
+      "step": 25251
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0560398505603984e-05,
+      "loss": 0.0921,
+      "step": 25252
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.054794520547945e-05,
+      "loss": 0.0987,
+      "step": 25253
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0535491905354918e-05,
+      "loss": 0.0671,
+      "step": 25254
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0523038605230384e-05,
+      "loss": 0.0652,
+      "step": 25255
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0510585305105853e-05,
+      "loss": 0.1499,
+      "step": 25256
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0498132004981319e-05,
+      "loss": 0.1232,
+      "step": 25257
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0485678704856787e-05,
+      "loss": 0.0898,
+      "step": 25258
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0473225404732253e-05,
+      "loss": 0.146,
+      "step": 25259
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.046077210460772e-05,
+      "loss": 0.0666,
+      "step": 25260
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0448318804483186e-05,
+      "loss": 0.0636,
+      "step": 25261
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0435865504358654e-05,
+      "loss": 0.0844,
+      "step": 25262
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.042341220423412e-05,
+      "loss": 0.189,
+      "step": 25263
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0410958904109589e-05,
+      "loss": 0.0951,
+      "step": 25264
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0398505603985055e-05,
+      "loss": 0.0893,
+      "step": 25265
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0386052303860523e-05,
+      "loss": 0.079,
+      "step": 25266
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0373599003735989e-05,
+      "loss": 0.0724,
+      "step": 25267
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0361145703611456e-05,
+      "loss": 0.1944,
+      "step": 25268
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0348692403486922e-05,
+      "loss": 0.2177,
+      "step": 25269
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.033623910336239e-05,
+      "loss": 0.0897,
+      "step": 25270
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0323785803237856e-05,
+      "loss": 0.1057,
+      "step": 25271
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0311332503113325e-05,
+      "loss": 0.0925,
+      "step": 25272
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0298879202988791e-05,
+      "loss": 0.0771,
+      "step": 25273
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0286425902864259e-05,
+      "loss": 0.0931,
+      "step": 25274
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0273972602739725e-05,
+      "loss": 0.1562,
+      "step": 25275
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0261519302615192e-05,
+      "loss": 0.2425,
+      "step": 25276
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0249066002490658e-05,
+      "loss": 0.0737,
+      "step": 25277
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0236612702366126e-05,
+      "loss": 0.1352,
+      "step": 25278
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0224159402241593e-05,
+      "loss": 0.0473,
+      "step": 25279
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0211706102117061e-05,
+      "loss": 0.1556,
+      "step": 25280
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0199252801992527e-05,
+      "loss": 0.1578,
+      "step": 25281
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0186799501867995e-05,
+      "loss": 0.156,
+      "step": 25282
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.017434620174346e-05,
+      "loss": 0.1574,
+      "step": 25283
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0161892901618928e-05,
+      "loss": 0.1048,
+      "step": 25284
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0149439601494394e-05,
+      "loss": 0.0892,
+      "step": 25285
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0136986301369864e-05,
+      "loss": 0.1428,
+      "step": 25286
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.012453300124533e-05,
+      "loss": 0.1183,
+      "step": 25287
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0112079701120797e-05,
+      "loss": 0.2155,
+      "step": 25288
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0099626400996263e-05,
+      "loss": 0.0912,
+      "step": 25289
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.008717310087173e-05,
+      "loss": 0.0735,
+      "step": 25290
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0074719800747197e-05,
+      "loss": 0.0868,
+      "step": 25291
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0062266500622664e-05,
+      "loss": 0.0684,
+      "step": 25292
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.004981320049813e-05,
+      "loss": 0.0723,
+      "step": 25293
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.00373599003736e-05,
+      "loss": 0.1637,
+      "step": 25294
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.0024906600249066e-05,
+      "loss": 0.1725,
+      "step": 25295
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.0012453300124533e-05,
+      "loss": 0.1638,
+      "step": 25296
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.999999999999999e-06,
+      "loss": 0.1337,
+      "step": 25297
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.987546699875467e-06,
+      "loss": 0.0764,
+      "step": 25298
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.975093399750933e-06,
+      "loss": 0.0949,
+      "step": 25299
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.9626400996264e-06,
+      "loss": 0.0901,
+      "step": 25300
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.950186799501866e-06,
+      "loss": 0.1782,
+      "step": 25301
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.937733499377336e-06,
+      "loss": 0.1513,
+      "step": 25302
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.925280199252802e-06,
+      "loss": 0.1322,
+      "step": 25303
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.91282689912827e-06,
+      "loss": 0.0706,
+      "step": 25304
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.900373599003735e-06,
+      "loss": 0.0878,
+      "step": 25305
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.887920298879201e-06,
+      "loss": 0.1953,
+      "step": 25306
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.875466998754669e-06,
+      "loss": 0.1386,
+      "step": 25307
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.863013698630135e-06,
+      "loss": 0.1704,
+      "step": 25308
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.850560398505604e-06,
+      "loss": 0.0883,
+      "step": 25309
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.83810709838107e-06,
+      "loss": 0.0684,
+      "step": 25310
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.825653798256538e-06,
+      "loss": 0.0585,
+      "step": 25311
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.813200498132004e-06,
+      "loss": 0.1106,
+      "step": 25312
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.800747198007471e-06,
+      "loss": 0.1964,
+      "step": 25313
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.788293897882937e-06,
+      "loss": 0.1152,
+      "step": 25314
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.775840597758405e-06,
+      "loss": 0.1242,
+      "step": 25315
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.76338729763387e-06,
+      "loss": 0.1597,
+      "step": 25316
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.75093399750934e-06,
+      "loss": 0.0748,
+      "step": 25317
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.738480697384806e-06,
+      "loss": 0.1075,
+      "step": 25318
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.726027397260274e-06,
+      "loss": 0.1328,
+      "step": 25319
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 9.71357409713574e-06,
+      "loss": 0.103,
+      "step": 25320
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.701120797011207e-06,
+      "loss": 0.1131,
+      "step": 25321
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.688667496886673e-06,
+      "loss": 0.072,
+      "step": 25322
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.67621419676214e-06,
+      "loss": 0.0858,
+      "step": 25323
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.663760896637607e-06,
+      "loss": 0.1353,
+      "step": 25324
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.651307596513076e-06,
+      "loss": 0.1288,
+      "step": 25325
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.638854296388542e-06,
+      "loss": 0.1349,
+      "step": 25326
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.62640099626401e-06,
+      "loss": 0.0917,
+      "step": 25327
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.613947696139476e-06,
+      "loss": 0.0765,
+      "step": 25328
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.601494396014943e-06,
+      "loss": 0.0694,
+      "step": 25329
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.58904109589041e-06,
+      "loss": 0.0665,
+      "step": 25330
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.576587795765877e-06,
+      "loss": 0.1983,
+      "step": 25331
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.564134495641344e-06,
+      "loss": 0.1925,
+      "step": 25332
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.551681195516812e-06,
+      "loss": 0.1187,
+      "step": 25333
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.539227895392278e-06,
+      "loss": 0.0851,
+      "step": 25334
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.526774595267746e-06,
+      "loss": 0.1172,
+      "step": 25335
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.514321295143212e-06,
+      "loss": 0.1935,
+      "step": 25336
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.50186799501868e-06,
+      "loss": 0.192,
+      "step": 25337
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.489414694894145e-06,
+      "loss": 0.2573,
+      "step": 25338
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.476961394769615e-06,
+      "loss": 0.1743,
+      "step": 25339
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.46450809464508e-06,
+      "loss": 0.1584,
+      "step": 25340
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.452054794520548e-06,
+      "loss": 0.0757,
+      "step": 25341
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.439601494396014e-06,
+      "loss": 0.103,
+      "step": 25342
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.427148194271482e-06,
+      "loss": 0.2503,
+      "step": 25343
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.414694894146948e-06,
+      "loss": 0.22,
+      "step": 25344
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.402241594022415e-06,
+      "loss": 0.2218,
+      "step": 25345
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 9.389788293897881e-06,
+      "loss": 0.116,
+      "step": 25346
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.37733499377335e-06,
+      "loss": 0.0759,
+      "step": 25347
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.364881693648817e-06,
+      "loss": 0.1149,
+      "step": 25348
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.352428393524282e-06,
+      "loss": 0.079,
+      "step": 25349
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.33997509339975e-06,
+      "loss": 0.2142,
+      "step": 25350
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.327521793275218e-06,
+      "loss": 0.1014,
+      "step": 25351
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.315068493150684e-06,
+      "loss": 0.1249,
+      "step": 25352
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.302615193026151e-06,
+      "loss": 0.1431,
+      "step": 25353
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.290161892901617e-06,
+      "loss": 0.0651,
+      "step": 25354
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.277708592777085e-06,
+      "loss": 0.0882,
+      "step": 25355
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.265255292652553e-06,
+      "loss": 0.1086,
+      "step": 25356
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.252801992528019e-06,
+      "loss": 0.2018,
+      "step": 25357
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.240348692403486e-06,
+      "loss": 0.1124,
+      "step": 25358
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.227895392278954e-06,
+      "loss": 0.093,
+      "step": 25359
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.21544209215442e-06,
+      "loss": 0.0726,
+      "step": 25360
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.202988792029887e-06,
+      "loss": 0.1522,
+      "step": 25361
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.190535491905355e-06,
+      "loss": 0.0652,
+      "step": 25362
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.178082191780821e-06,
+      "loss": 0.1394,
+      "step": 25363
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.165628891656289e-06,
+      "loss": 0.1268,
+      "step": 25364
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.153175591531755e-06,
+      "loss": 0.0999,
+      "step": 25365
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.140722291407222e-06,
+      "loss": 0.1438,
+      "step": 25366
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.12826899128269e-06,
+      "loss": 0.1244,
+      "step": 25367
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.115815691158156e-06,
+      "loss": 0.0627,
+      "step": 25368
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.103362391033623e-06,
+      "loss": 0.1425,
+      "step": 25369
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.141,
+      "step": 25370
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.078455790784557e-06,
+      "loss": 0.1282,
+      "step": 25371
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 9.066002490660025e-06,
+      "loss": 0.0881,
+      "step": 25372
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 9.053549190535492e-06,
+      "loss": 0.1,
+      "step": 25373
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 9.041095890410958e-06,
+      "loss": 0.0906,
+      "step": 25374
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 9.028642590286426e-06,
+      "loss": 0.2095,
+      "step": 25375
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 9.016189290161892e-06,
+      "loss": 0.134,
+      "step": 25376
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 9.00373599003736e-06,
+      "loss": 0.0966,
+      "step": 25377
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.991282689912827e-06,
+      "loss": 0.065,
+      "step": 25378
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.978829389788293e-06,
+      "loss": 0.1326,
+      "step": 25379
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.96637608966376e-06,
+      "loss": 0.0728,
+      "step": 25380
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.953922789539228e-06,
+      "loss": 0.1503,
+      "step": 25381
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.941469489414694e-06,
+      "loss": 0.1292,
+      "step": 25382
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.929016189290162e-06,
+      "loss": 0.1144,
+      "step": 25383
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.916562889165628e-06,
+      "loss": 0.0989,
+      "step": 25384
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.904109589041095e-06,
+      "loss": 0.1098,
+      "step": 25385
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.891656288916561e-06,
+      "loss": 0.1206,
+      "step": 25386
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.879202988792029e-06,
+      "loss": 0.1103,
+      "step": 25387
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.866749688667495e-06,
+      "loss": 0.179,
+      "step": 25388
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.854296388542963e-06,
+      "loss": 0.1708,
+      "step": 25389
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.84184308841843e-06,
+      "loss": 0.0845,
+      "step": 25390
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.829389788293896e-06,
+      "loss": 0.0666,
+      "step": 25391
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.816936488169364e-06,
+      "loss": 0.0644,
+      "step": 25392
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.804483188044831e-06,
+      "loss": 0.1426,
+      "step": 25393
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.792029887920297e-06,
+      "loss": 0.1566,
+      "step": 25394
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.779576587795765e-06,
+      "loss": 0.1178,
+      "step": 25395
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.767123287671231e-06,
+      "loss": 0.0919,
+      "step": 25396
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.754669987546699e-06,
+      "loss": 0.1223,
+      "step": 25397
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 8.742216687422166e-06,
+      "loss": 0.1469,
+      "step": 25398
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.729763387297632e-06,
+      "loss": 0.0776,
+      "step": 25399
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.7173100871731e-06,
+      "loss": 0.2131,
+      "step": 25400
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.704856787048568e-06,
+      "loss": 0.1767,
+      "step": 25401
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.692403486924033e-06,
+      "loss": 0.1398,
+      "step": 25402
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.679950186799501e-06,
+      "loss": 0.0754,
+      "step": 25403
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.667496886674969e-06,
+      "loss": 0.0751,
+      "step": 25404
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.655043586550435e-06,
+      "loss": 0.0984,
+      "step": 25405
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.642590286425902e-06,
+      "loss": 0.1749,
+      "step": 25406
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.630136986301368e-06,
+      "loss": 0.1507,
+      "step": 25407
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.617683686176836e-06,
+      "loss": 0.1441,
+      "step": 25408
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.605230386052304e-06,
+      "loss": 0.1371,
+      "step": 25409
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.59277708592777e-06,
+      "loss": 0.2309,
+      "step": 25410
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.580323785803237e-06,
+      "loss": 0.0981,
+      "step": 25411
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.567870485678705e-06,
+      "loss": 0.0556,
+      "step": 25412
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.55541718555417e-06,
+      "loss": 0.2334,
+      "step": 25413
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.542963885429638e-06,
+      "loss": 0.1653,
+      "step": 25414
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.530510585305106e-06,
+      "loss": 0.1385,
+      "step": 25415
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.518057285180572e-06,
+      "loss": 0.113,
+      "step": 25416
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.50560398505604e-06,
+      "loss": 0.102,
+      "step": 25417
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.493150684931506e-06,
+      "loss": 0.1046,
+      "step": 25418
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.480697384806973e-06,
+      "loss": 0.1498,
+      "step": 25419
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.46824408468244e-06,
+      "loss": 0.1046,
+      "step": 25420
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.455790784557907e-06,
+      "loss": 0.1517,
+      "step": 25421
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.443337484433374e-06,
+      "loss": 0.076,
+      "step": 25422
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.430884184308842e-06,
+      "loss": 0.0636,
+      "step": 25423
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 8.418430884184308e-06,
+      "loss": 0.1266,
+      "step": 25424
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.405977584059776e-06,
+      "loss": 0.1616,
+      "step": 25425
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.393524283935243e-06,
+      "loss": 0.1737,
+      "step": 25426
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.38107098381071e-06,
+      "loss": 0.2169,
+      "step": 25427
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.368617683686177e-06,
+      "loss": 0.1272,
+      "step": 25428
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.356164383561643e-06,
+      "loss": 0.1443,
+      "step": 25429
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.34371108343711e-06,
+      "loss": 0.0851,
+      "step": 25430
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.331257783312578e-06,
+      "loss": 0.1542,
+      "step": 25431
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.318804483188044e-06,
+      "loss": 0.1138,
+      "step": 25432
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.306351183063512e-06,
+      "loss": 0.1439,
+      "step": 25433
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.29389788293898e-06,
+      "loss": 0.1638,
+      "step": 25434
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.281444582814445e-06,
+      "loss": 0.1184,
+      "step": 25435
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.268991282689913e-06,
+      "loss": 0.063,
+      "step": 25436
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.256537982565379e-06,
+      "loss": 0.1341,
+      "step": 25437
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.244084682440846e-06,
+      "loss": 0.1455,
+      "step": 25438
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.231631382316312e-06,
+      "loss": 0.1129,
+      "step": 25439
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.21917808219178e-06,
+      "loss": 0.0785,
+      "step": 25440
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.206724782067246e-06,
+      "loss": 0.1722,
+      "step": 25441
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.194271481942714e-06,
+      "loss": 0.0572,
+      "step": 25442
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.181818181818181e-06,
+      "loss": 0.0767,
+      "step": 25443
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.169364881693647e-06,
+      "loss": 0.183,
+      "step": 25444
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.156911581569115e-06,
+      "loss": 0.1173,
+      "step": 25445
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.144458281444582e-06,
+      "loss": 0.173,
+      "step": 25446
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.132004981320048e-06,
+      "loss": 0.1137,
+      "step": 25447
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.119551681195516e-06,
+      "loss": 0.1025,
+      "step": 25448
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.107098381070982e-06,
+      "loss": 0.0672,
+      "step": 25449
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 8.09464508094645e-06,
+      "loss": 0.173,
+      "step": 25450
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 8.082191780821917e-06,
+      "loss": 0.156,
+      "step": 25451
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 8.069738480697383e-06,
+      "loss": 0.1073,
+      "step": 25452
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 8.057285180572851e-06,
+      "loss": 0.0955,
+      "step": 25453
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 8.044831880448319e-06,
+      "loss": 0.0881,
+      "step": 25454
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 8.032378580323784e-06,
+      "loss": 0.1351,
+      "step": 25455
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 8.019925280199252e-06,
+      "loss": 0.1286,
+      "step": 25456
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 8.00747198007472e-06,
+      "loss": 0.1318,
+      "step": 25457
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.995018679950186e-06,
+      "loss": 0.1077,
+      "step": 25458
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.982565379825653e-06,
+      "loss": 0.0865,
+      "step": 25459
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.97011207970112e-06,
+      "loss": 0.1024,
+      "step": 25460
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.957658779576587e-06,
+      "loss": 0.0593,
+      "step": 25461
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.945205479452055e-06,
+      "loss": 0.0752,
+      "step": 25462
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.93275217932752e-06,
+      "loss": 0.1966,
+      "step": 25463
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.920298879202988e-06,
+      "loss": 0.1677,
+      "step": 25464
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.907845579078456e-06,
+      "loss": 0.1207,
+      "step": 25465
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.895392278953922e-06,
+      "loss": 0.1158,
+      "step": 25466
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.88293897882939e-06,
+      "loss": 0.146,
+      "step": 25467
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.870485678704857e-06,
+      "loss": 0.0366,
+      "step": 25468
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.858032378580323e-06,
+      "loss": 0.1575,
+      "step": 25469
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.84557907845579e-06,
+      "loss": 0.2154,
+      "step": 25470
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.833125778331257e-06,
+      "loss": 0.1017,
+      "step": 25471
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.820672478206724e-06,
+      "loss": 0.0659,
+      "step": 25472
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.808219178082192e-06,
+      "loss": 0.1013,
+      "step": 25473
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.795765877957658e-06,
+      "loss": 0.1519,
+      "step": 25474
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.783312577833125e-06,
+      "loss": 0.0981,
+      "step": 25475
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 7.770859277708593e-06,
+      "loss": 0.0997,
+      "step": 25476
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.758405977584059e-06,
+      "loss": 0.2753,
+      "step": 25477
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.745952677459527e-06,
+      "loss": 0.0953,
+      "step": 25478
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.733499377334994e-06,
+      "loss": 0.0864,
+      "step": 25479
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.72104607721046e-06,
+      "loss": 0.0902,
+      "step": 25480
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.708592777085928e-06,
+      "loss": 0.076,
+      "step": 25481
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.696139476961394e-06,
+      "loss": 0.1509,
+      "step": 25482
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.683686176836861e-06,
+      "loss": 0.141,
+      "step": 25483
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.671232876712329e-06,
+      "loss": 0.098,
+      "step": 25484
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.658779576587795e-06,
+      "loss": 0.1739,
+      "step": 25485
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.646326276463263e-06,
+      "loss": 0.0578,
+      "step": 25486
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.63387297633873e-06,
+      "loss": 0.1165,
+      "step": 25487
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.621419676214196e-06,
+      "loss": 0.2267,
+      "step": 25488
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.608966376089664e-06,
+      "loss": 0.099,
+      "step": 25489
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.59651307596513e-06,
+      "loss": 0.1849,
+      "step": 25490
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.584059775840597e-06,
+      "loss": 0.1185,
+      "step": 25491
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.571606475716063e-06,
+      "loss": 0.0931,
+      "step": 25492
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.559153175591531e-06,
+      "loss": 0.0603,
+      "step": 25493
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.546699875466998e-06,
+      "loss": 0.0951,
+      "step": 25494
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.534246575342465e-06,
+      "loss": 0.1767,
+      "step": 25495
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.521793275217931e-06,
+      "loss": 0.1411,
+      "step": 25496
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.509339975093399e-06,
+      "loss": 0.1546,
+      "step": 25497
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.496886674968866e-06,
+      "loss": 0.1188,
+      "step": 25498
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.484433374844333e-06,
+      "loss": 0.0692,
+      "step": 25499
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.471980074719799e-06,
+      "loss": 0.1823,
+      "step": 25500
+    },
+    {
+      "epoch": 9.77,
+      "eval_cer": 0.09994732426790487,
+      "eval_loss": 0.42302536964416504,
+      "eval_runtime": 47.755,
+      "eval_samples_per_second": 36.708,
+      "eval_steps_per_second": 4.607,
+      "eval_wer": 0.15624292890437616,
+      "step": 25500
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.459526774595267e-06,
+      "loss": 0.1185,
+      "step": 25501
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 7.447073474470734e-06,
+      "loss": 0.0782,
+      "step": 25502
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.434620174346201e-06,
+      "loss": 0.0993,
+      "step": 25503
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.4221668742216674e-06,
+      "loss": 0.0742,
+      "step": 25504
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.409713574097135e-06,
+      "loss": 0.0424,
+      "step": 25505
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.397260273972602e-06,
+      "loss": 0.1506,
+      "step": 25506
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.384806973848069e-06,
+      "loss": 0.1703,
+      "step": 25507
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.372353673723536e-06,
+      "loss": 0.1202,
+      "step": 25508
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.359900373599003e-06,
+      "loss": 0.1061,
+      "step": 25509
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.34744707347447e-06,
+      "loss": 0.1763,
+      "step": 25510
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.334993773349937e-06,
+      "loss": 0.1132,
+      "step": 25511
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.322540473225404e-06,
+      "loss": 0.1278,
+      "step": 25512
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.310087173100871e-06,
+      "loss": 0.1993,
+      "step": 25513
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.297633872976338e-06,
+      "loss": 0.0904,
+      "step": 25514
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.285180572851805e-06,
+      "loss": 0.1302,
+      "step": 25515
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.272727272727272e-06,
+      "loss": 0.0913,
+      "step": 25516
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.260273972602739e-06,
+      "loss": 0.0574,
+      "step": 25517
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.247820672478206e-06,
+      "loss": 0.0886,
+      "step": 25518
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.2353673723536735e-06,
+      "loss": 0.0726,
+      "step": 25519
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.22291407222914e-06,
+      "loss": 0.1372,
+      "step": 25520
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.210460772104607e-06,
+      "loss": 0.0917,
+      "step": 25521
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.198007471980074e-06,
+      "loss": 0.0665,
+      "step": 25522
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.1855541718555416e-06,
+      "loss": 0.0692,
+      "step": 25523
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.173100871731008e-06,
+      "loss": 0.1154,
+      "step": 25524
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.160647571606475e-06,
+      "loss": 0.2166,
+      "step": 25525
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.148194271481942e-06,
+      "loss": 0.1262,
+      "step": 25526
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.1357409713574096e-06,
+      "loss": 0.1364,
+      "step": 25527
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.123287671232876e-06,
+      "loss": 0.2103,
+      "step": 25528
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 7.110834371108343e-06,
+      "loss": 0.0813,
+      "step": 25529
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.09838107098381e-06,
+      "loss": 0.0629,
+      "step": 25530
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.085927770859278e-06,
+      "loss": 0.1367,
+      "step": 25531
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.073474470734744e-06,
+      "loss": 0.2103,
+      "step": 25532
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.061021170610211e-06,
+      "loss": 0.17,
+      "step": 25533
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.048567870485679e-06,
+      "loss": 0.22,
+      "step": 25534
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.036114570361146e-06,
+      "loss": 0.1042,
+      "step": 25535
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.023661270236612e-06,
+      "loss": 0.0968,
+      "step": 25536
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 7.011207970112079e-06,
+      "loss": 0.1079,
+      "step": 25537
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.998754669987547e-06,
+      "loss": 0.2253,
+      "step": 25538
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.986301369863014e-06,
+      "loss": 0.1046,
+      "step": 25539
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.97384806973848e-06,
+      "loss": 0.0695,
+      "step": 25540
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.961394769613947e-06,
+      "loss": 0.065,
+      "step": 25541
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.948941469489415e-06,
+      "loss": 0.0572,
+      "step": 25542
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.936488169364881e-06,
+      "loss": 0.1124,
+      "step": 25543
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.924034869240348e-06,
+      "loss": 0.118,
+      "step": 25544
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.911581569115814e-06,
+      "loss": 0.1339,
+      "step": 25545
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.899128268991282e-06,
+      "loss": 0.0714,
+      "step": 25546
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.886674968866749e-06,
+      "loss": 0.0542,
+      "step": 25547
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.874221668742216e-06,
+      "loss": 0.0538,
+      "step": 25548
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.861768368617682e-06,
+      "loss": 0.1199,
+      "step": 25549
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.84931506849315e-06,
+      "loss": 0.1172,
+      "step": 25550
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.836861768368617e-06,
+      "loss": 0.0957,
+      "step": 25551
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.824408468244084e-06,
+      "loss": 0.1325,
+      "step": 25552
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.81195516811955e-06,
+      "loss": 0.0773,
+      "step": 25553
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.799501867995018e-06,
+      "loss": 0.1516,
+      "step": 25554
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.787048567870485e-06,
+      "loss": 0.1271,
+      "step": 25555
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.774595267745952e-06,
+      "loss": 0.158,
+      "step": 25556
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.762141967621418e-06,
+      "loss": 0.1565,
+      "step": 25557
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.749688667496886e-06,
+      "loss": 0.1177,
+      "step": 25558
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.737235367372353e-06,
+      "loss": 0.1673,
+      "step": 25559
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.72478206724782e-06,
+      "loss": 0.0657,
+      "step": 25560
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.712328767123287e-06,
+      "loss": 0.1108,
+      "step": 25561
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.699875466998754e-06,
+      "loss": 0.192,
+      "step": 25562
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.687422166874221e-06,
+      "loss": 0.1791,
+      "step": 25563
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.674968866749688e-06,
+      "loss": 0.1855,
+      "step": 25564
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.662515566625155e-06,
+      "loss": 0.2147,
+      "step": 25565
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.650062266500622e-06,
+      "loss": 0.1502,
+      "step": 25566
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.637608966376089e-06,
+      "loss": 0.0942,
+      "step": 25567
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.625155666251556e-06,
+      "loss": 0.1194,
+      "step": 25568
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.612702366127023e-06,
+      "loss": 0.2545,
+      "step": 25569
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.60024906600249e-06,
+      "loss": 0.09,
+      "step": 25570
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.587795765877957e-06,
+      "loss": 0.0849,
+      "step": 25571
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.5753424657534245e-06,
+      "loss": 0.0836,
+      "step": 25572
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.562889165628891e-06,
+      "loss": 0.0706,
+      "step": 25573
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.550435865504358e-06,
+      "loss": 0.04,
+      "step": 25574
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.537982565379825e-06,
+      "loss": 0.1091,
+      "step": 25575
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.5255292652552925e-06,
+      "loss": 0.1377,
+      "step": 25576
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.513075965130759e-06,
+      "loss": 0.1648,
+      "step": 25577
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.500622665006226e-06,
+      "loss": 0.0666,
+      "step": 25578
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.488169364881693e-06,
+      "loss": 0.0897,
+      "step": 25579
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.4757160647571606e-06,
+      "loss": 0.084,
+      "step": 25580
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 6.463262764632627e-06,
+      "loss": 0.2333,
+      "step": 25581
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.450809464508094e-06,
+      "loss": 0.1342,
+      "step": 25582
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.438356164383561e-06,
+      "loss": 0.1032,
+      "step": 25583
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.425902864259029e-06,
+      "loss": 0.0971,
+      "step": 25584
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.413449564134495e-06,
+      "loss": 0.0727,
+      "step": 25585
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.400996264009962e-06,
+      "loss": 0.103,
+      "step": 25586
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.38854296388543e-06,
+      "loss": 0.106,
+      "step": 25587
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.376089663760897e-06,
+      "loss": 0.1064,
+      "step": 25588
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.363636363636363e-06,
+      "loss": 0.1358,
+      "step": 25589
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.35118306351183e-06,
+      "loss": 0.1058,
+      "step": 25590
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.338729763387298e-06,
+      "loss": 0.0754,
+      "step": 25591
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.326276463262765e-06,
+      "loss": 0.1043,
+      "step": 25592
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.313823163138231e-06,
+      "loss": 0.0572,
+      "step": 25593
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.301369863013698e-06,
+      "loss": 0.1394,
+      "step": 25594
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.288916562889166e-06,
+      "loss": 0.1343,
+      "step": 25595
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.276463262764632e-06,
+      "loss": 0.1798,
+      "step": 25596
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.2640099626400986e-06,
+      "loss": 0.1066,
+      "step": 25597
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.251556662515565e-06,
+      "loss": 0.1186,
+      "step": 25598
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.239103362391033e-06,
+      "loss": 0.1198,
+      "step": 25599
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.2266500622665e-06,
+      "loss": 0.178,
+      "step": 25600
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.214196762141967e-06,
+      "loss": 0.1175,
+      "step": 25601
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.201743462017433e-06,
+      "loss": 0.0935,
+      "step": 25602
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.189290161892901e-06,
+      "loss": 0.1417,
+      "step": 25603
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.176836861768368e-06,
+      "loss": 0.1227,
+      "step": 25604
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.164383561643835e-06,
+      "loss": 0.0605,
+      "step": 25605
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.151930261519301e-06,
+      "loss": 0.1445,
+      "step": 25606
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.139476961394769e-06,
+      "loss": 0.1417,
+      "step": 25607
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.127023661270236e-06,
+      "loss": 0.1528,
+      "step": 25608
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.114570361145703e-06,
+      "loss": 0.163,
+      "step": 25609
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.10211706102117e-06,
+      "loss": 0.0785,
+      "step": 25610
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.089663760896637e-06,
+      "loss": 0.0649,
+      "step": 25611
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.077210460772104e-06,
+      "loss": 0.0636,
+      "step": 25612
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.064757160647571e-06,
+      "loss": 0.2078,
+      "step": 25613
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.052303860523038e-06,
+      "loss": 0.1332,
+      "step": 25614
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.039850560398505e-06,
+      "loss": 0.0743,
+      "step": 25615
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.027397260273972e-06,
+      "loss": 0.1536,
+      "step": 25616
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.014943960149439e-06,
+      "loss": 0.1541,
+      "step": 25617
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.002490660024906e-06,
+      "loss": 0.1129,
+      "step": 25618
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.990037359900373e-06,
+      "loss": 0.2556,
+      "step": 25619
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.97758405977584e-06,
+      "loss": 0.1349,
+      "step": 25620
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.965130759651307e-06,
+      "loss": 0.0877,
+      "step": 25621
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.952677459526774e-06,
+      "loss": 0.0945,
+      "step": 25622
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.940224159402241e-06,
+      "loss": 0.118,
+      "step": 25623
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.927770859277708e-06,
+      "loss": 0.096,
+      "step": 25624
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.9153175591531755e-06,
+      "loss": 0.1698,
+      "step": 25625
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.902864259028642e-06,
+      "loss": 0.1354,
+      "step": 25626
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.890410958904109e-06,
+      "loss": 0.1234,
+      "step": 25627
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.877957658779576e-06,
+      "loss": 0.1192,
+      "step": 25628
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.8655043586550435e-06,
+      "loss": 0.0981,
+      "step": 25629
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.85305105853051e-06,
+      "loss": 0.1493,
+      "step": 25630
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.840597758405977e-06,
+      "loss": 0.1416,
+      "step": 25631
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.828144458281444e-06,
+      "loss": 0.2558,
+      "step": 25632
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 5.8156911581569116e-06,
+      "loss": 0.2424,
+      "step": 25633
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.803237858032378e-06,
+      "loss": 0.084,
+      "step": 25634
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.790784557907845e-06,
+      "loss": 0.1466,
+      "step": 25635
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.778331257783312e-06,
+      "loss": 0.1203,
+      "step": 25636
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.7658779576587796e-06,
+      "loss": 0.0936,
+      "step": 25637
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.753424657534246e-06,
+      "loss": 0.1204,
+      "step": 25638
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.740971357409713e-06,
+      "loss": 0.1427,
+      "step": 25639
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.728518057285181e-06,
+      "loss": 0.1717,
+      "step": 25640
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.716064757160648e-06,
+      "loss": 0.088,
+      "step": 25641
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.703611457036114e-06,
+      "loss": 0.0541,
+      "step": 25642
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.691158156911581e-06,
+      "loss": 0.1186,
+      "step": 25643
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.678704856787049e-06,
+      "loss": 0.1429,
+      "step": 25644
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.666251556662516e-06,
+      "loss": 0.1934,
+      "step": 25645
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.653798256537982e-06,
+      "loss": 0.1521,
+      "step": 25646
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.641344956413449e-06,
+      "loss": 0.1097,
+      "step": 25647
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.628891656288917e-06,
+      "loss": 0.1997,
+      "step": 25648
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.616438356164383e-06,
+      "loss": 0.0947,
+      "step": 25649
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.6039850560398496e-06,
+      "loss": 0.1185,
+      "step": 25650
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.591531755915316e-06,
+      "loss": 0.1684,
+      "step": 25651
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.579078455790784e-06,
+      "loss": 0.1588,
+      "step": 25652
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.566625155666251e-06,
+      "loss": 0.1512,
+      "step": 25653
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.5541718555417176e-06,
+      "loss": 0.1583,
+      "step": 25654
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.541718555417184e-06,
+      "loss": 0.0908,
+      "step": 25655
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.529265255292652e-06,
+      "loss": 0.0821,
+      "step": 25656
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.516811955168119e-06,
+      "loss": 0.1847,
+      "step": 25657
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.504358655043586e-06,
+      "loss": 0.1114,
+      "step": 25658
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 5.491905354919052e-06,
+      "loss": 0.0948,
+      "step": 25659
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.47945205479452e-06,
+      "loss": 0.0834,
+      "step": 25660
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.466998754669987e-06,
+      "loss": 0.0592,
+      "step": 25661
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.454545454545454e-06,
+      "loss": 0.0897,
+      "step": 25662
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.442092154420921e-06,
+      "loss": 0.1327,
+      "step": 25663
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.429638854296388e-06,
+      "loss": 0.1396,
+      "step": 25664
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.417185554171855e-06,
+      "loss": 0.0745,
+      "step": 25665
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.404732254047322e-06,
+      "loss": 0.1265,
+      "step": 25666
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.392278953922789e-06,
+      "loss": 0.1304,
+      "step": 25667
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.379825653798256e-06,
+      "loss": 0.1104,
+      "step": 25668
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.367372353673723e-06,
+      "loss": 0.1601,
+      "step": 25669
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.35491905354919e-06,
+      "loss": 0.0856,
+      "step": 25670
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.342465753424657e-06,
+      "loss": 0.1145,
+      "step": 25671
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.330012453300124e-06,
+      "loss": 0.0988,
+      "step": 25672
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.317559153175591e-06,
+      "loss": 0.1611,
+      "step": 25673
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.305105853051058e-06,
+      "loss": 0.1182,
+      "step": 25674
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.292652552926525e-06,
+      "loss": 0.1263,
+      "step": 25675
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.280199252801992e-06,
+      "loss": 0.1224,
+      "step": 25676
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.267745952677459e-06,
+      "loss": 0.0875,
+      "step": 25677
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.2552926525529265e-06,
+      "loss": 0.1043,
+      "step": 25678
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.242839352428393e-06,
+      "loss": 0.0946,
+      "step": 25679
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.23038605230386e-06,
+      "loss": 0.0543,
+      "step": 25680
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.217932752179327e-06,
+      "loss": 0.2128,
+      "step": 25681
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.2054794520547945e-06,
+      "loss": 0.1378,
+      "step": 25682
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.193026151930261e-06,
+      "loss": 0.1057,
+      "step": 25683
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.180572851805728e-06,
+      "loss": 0.124,
+      "step": 25684
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 5.168119551681195e-06,
+      "loss": 0.1257,
+      "step": 25685
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.1556662515566625e-06,
+      "loss": 0.1065,
+      "step": 25686
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.143212951432129e-06,
+      "loss": 0.0876,
+      "step": 25687
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.130759651307596e-06,
+      "loss": 0.1161,
+      "step": 25688
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.118306351183063e-06,
+      "loss": 0.1201,
+      "step": 25689
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.1058530510585306e-06,
+      "loss": 0.1062,
+      "step": 25690
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.093399750933997e-06,
+      "loss": 0.0711,
+      "step": 25691
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.080946450809464e-06,
+      "loss": 0.0929,
+      "step": 25692
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.068493150684932e-06,
+      "loss": 0.0568,
+      "step": 25693
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.0560398505603986e-06,
+      "loss": 0.1717,
+      "step": 25694
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.043586550435865e-06,
+      "loss": 0.12,
+      "step": 25695
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.031133250311332e-06,
+      "loss": 0.1648,
+      "step": 25696
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.0186799501868e-06,
+      "loss": 0.117,
+      "step": 25697
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 5.006226650062267e-06,
+      "loss": 0.0844,
+      "step": 25698
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.993773349937733e-06,
+      "loss": 0.1498,
+      "step": 25699
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.9813200498132e-06,
+      "loss": 0.0678,
+      "step": 25700
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.968866749688668e-06,
+      "loss": 0.168,
+      "step": 25701
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.956413449564135e-06,
+      "loss": 0.1264,
+      "step": 25702
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.9439601494396005e-06,
+      "loss": 0.0934,
+      "step": 25703
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.931506849315067e-06,
+      "loss": 0.066,
+      "step": 25704
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.919053549190535e-06,
+      "loss": 0.1028,
+      "step": 25705
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.906600249066002e-06,
+      "loss": 0.1622,
+      "step": 25706
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.8941469489414686e-06,
+      "loss": 0.1543,
+      "step": 25707
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.881693648816935e-06,
+      "loss": 0.1546,
+      "step": 25708
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.869240348692403e-06,
+      "loss": 0.0902,
+      "step": 25709
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.85678704856787e-06,
+      "loss": 0.0626,
+      "step": 25710
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 4.844333748443337e-06,
+      "loss": 0.1636,
+      "step": 25711
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.831880448318803e-06,
+      "loss": 0.1085,
+      "step": 25712
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.819427148194271e-06,
+      "loss": 0.1952,
+      "step": 25713
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.806973848069738e-06,
+      "loss": 0.1014,
+      "step": 25714
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.794520547945205e-06,
+      "loss": 0.0793,
+      "step": 25715
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.782067247820672e-06,
+      "loss": 0.1046,
+      "step": 25716
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.769613947696139e-06,
+      "loss": 0.1651,
+      "step": 25717
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.757160647571606e-06,
+      "loss": 0.1579,
+      "step": 25718
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.744707347447073e-06,
+      "loss": 0.1774,
+      "step": 25719
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.73225404732254e-06,
+      "loss": 0.1047,
+      "step": 25720
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.719800747198007e-06,
+      "loss": 0.1543,
+      "step": 25721
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.707347447073474e-06,
+      "loss": 0.0978,
+      "step": 25722
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.694894146948941e-06,
+      "loss": 0.2027,
+      "step": 25723
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.682440846824408e-06,
+      "loss": 0.1415,
+      "step": 25724
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.669987546699875e-06,
+      "loss": 0.1755,
+      "step": 25725
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.657534246575342e-06,
+      "loss": 0.1744,
+      "step": 25726
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.645080946450809e-06,
+      "loss": 0.1195,
+      "step": 25727
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.632627646326276e-06,
+      "loss": 0.0966,
+      "step": 25728
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.620174346201743e-06,
+      "loss": 0.0811,
+      "step": 25729
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.60772104607721e-06,
+      "loss": 0.1173,
+      "step": 25730
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.5952677459526775e-06,
+      "loss": 0.0875,
+      "step": 25731
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.582814445828144e-06,
+      "loss": 0.1291,
+      "step": 25732
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.570361145703611e-06,
+      "loss": 0.1787,
+      "step": 25733
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.557907845579078e-06,
+      "loss": 0.1211,
+      "step": 25734
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.0581,
+      "step": 25735
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.533001245330012e-06,
+      "loss": 0.1263,
+      "step": 25736
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 4.520547945205479e-06,
+      "loss": 0.1343,
+      "step": 25737
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.508094645080946e-06,
+      "loss": 0.1319,
+      "step": 25738
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.4956413449564135e-06,
+      "loss": 0.2278,
+      "step": 25739
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.48318804483188e-06,
+      "loss": 0.0808,
+      "step": 25740
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.470734744707347e-06,
+      "loss": 0.0729,
+      "step": 25741
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.458281444582814e-06,
+      "loss": 0.1592,
+      "step": 25742
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.445828144458281e-06,
+      "loss": 0.1562,
+      "step": 25743
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.4333748443337475e-06,
+      "loss": 0.1744,
+      "step": 25744
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.420921544209215e-06,
+      "loss": 0.1131,
+      "step": 25745
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.408468244084682e-06,
+      "loss": 0.0559,
+      "step": 25746
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.396014943960149e-06,
+      "loss": 0.0597,
+      "step": 25747
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.3835616438356155e-06,
+      "loss": 0.0921,
+      "step": 25748
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.371108343711083e-06,
+      "loss": 0.1138,
+      "step": 25749
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.35865504358655e-06,
+      "loss": 0.1404,
+      "step": 25750
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.346201743462017e-06,
+      "loss": 0.1674,
+      "step": 25751
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.333748443337484e-06,
+      "loss": 0.1603,
+      "step": 25752
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.321295143212951e-06,
+      "loss": 0.1277,
+      "step": 25753
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.308841843088418e-06,
+      "loss": 0.1266,
+      "step": 25754
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.296388542963885e-06,
+      "loss": 0.1161,
+      "step": 25755
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.283935242839352e-06,
+      "loss": 0.2018,
+      "step": 25756
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.271481942714819e-06,
+      "loss": 0.1381,
+      "step": 25757
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.259028642590286e-06,
+      "loss": 0.0964,
+      "step": 25758
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.246575342465753e-06,
+      "loss": 0.087,
+      "step": 25759
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.23412204234122e-06,
+      "loss": 0.0588,
+      "step": 25760
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.221668742216687e-06,
+      "loss": 0.129,
+      "step": 25761
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.209215442092154e-06,
+      "loss": 0.1378,
+      "step": 25762
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 4.196762141967622e-06,
+      "loss": 0.1735,
+      "step": 25763
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.184308841843088e-06,
+      "loss": 0.1379,
+      "step": 25764
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.171855541718555e-06,
+      "loss": 0.1092,
+      "step": 25765
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.159402241594022e-06,
+      "loss": 0.0838,
+      "step": 25766
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.14694894146949e-06,
+      "loss": 0.0675,
+      "step": 25767
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.1344956413449564e-06,
+      "loss": 0.0833,
+      "step": 25768
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.122042341220423e-06,
+      "loss": 0.1757,
+      "step": 25769
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.10958904109589e-06,
+      "loss": 0.1421,
+      "step": 25770
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.097135740971357e-06,
+      "loss": 0.0654,
+      "step": 25771
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.084682440846824e-06,
+      "loss": 0.104,
+      "step": 25772
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.072229140722291e-06,
+      "loss": 0.0634,
+      "step": 25773
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.059775840597758e-06,
+      "loss": 0.0819,
+      "step": 25774
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.047322540473225e-06,
+      "loss": 0.1897,
+      "step": 25775
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.034869240348692e-06,
+      "loss": 0.1581,
+      "step": 25776
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.022415940224159e-06,
+      "loss": 0.1909,
+      "step": 25777
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 4.009962640099626e-06,
+      "loss": 0.1021,
+      "step": 25778
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.997509339975093e-06,
+      "loss": 0.17,
+      "step": 25779
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.98505603985056e-06,
+      "loss": 0.1277,
+      "step": 25780
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.972602739726027e-06,
+      "loss": 0.1867,
+      "step": 25781
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.960149439601494e-06,
+      "loss": 0.1697,
+      "step": 25782
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.947696139476961e-06,
+      "loss": 0.1186,
+      "step": 25783
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.9352428393524285e-06,
+      "loss": 0.0713,
+      "step": 25784
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.922789539227895e-06,
+      "loss": 0.1073,
+      "step": 25785
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.910336239103362e-06,
+      "loss": 0.1557,
+      "step": 25786
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.897882938978829e-06,
+      "loss": 0.0955,
+      "step": 25787
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.8854296388542965e-06,
+      "loss": 0.1201,
+      "step": 25788
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.872976338729763e-06,
+      "loss": 0.2011,
+      "step": 25789
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.86052303860523e-06,
+      "loss": 0.1398,
+      "step": 25790
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.848069738480697e-06,
+      "loss": 0.1338,
+      "step": 25791
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.8356164383561645e-06,
+      "loss": 0.1173,
+      "step": 25792
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.823163138231631e-06,
+      "loss": 0.1418,
+      "step": 25793
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.810709838107098e-06,
+      "loss": 0.1141,
+      "step": 25794
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.798256537982565e-06,
+      "loss": 0.1452,
+      "step": 25795
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.7858032378580317e-06,
+      "loss": 0.0927,
+      "step": 25796
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.773349937733499e-06,
+      "loss": 0.0967,
+      "step": 25797
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.7608966376089657e-06,
+      "loss": 0.1345,
+      "step": 25798
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.748443337484433e-06,
+      "loss": 0.0998,
+      "step": 25799
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.7359900373598997e-06,
+      "loss": 0.1296,
+      "step": 25800
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.723536737235367e-06,
+      "loss": 0.1077,
+      "step": 25801
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.7110834371108337e-06,
+      "loss": 0.1506,
+      "step": 25802
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.698630136986301e-06,
+      "loss": 0.1516,
+      "step": 25803
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.686176836861768e-06,
+      "loss": 0.0816,
+      "step": 25804
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.673723536737235e-06,
+      "loss": 0.0927,
+      "step": 25805
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.661270236612702e-06,
+      "loss": 0.1459,
+      "step": 25806
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.648816936488169e-06,
+      "loss": 0.1209,
+      "step": 25807
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.636363636363636e-06,
+      "loss": 0.1119,
+      "step": 25808
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.623910336239103e-06,
+      "loss": 0.2069,
+      "step": 25809
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.61145703611457e-06,
+      "loss": 0.0663,
+      "step": 25810
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.599003735990037e-06,
+      "loss": 0.1247,
+      "step": 25811
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.586550435865504e-06,
+      "loss": 0.1339,
+      "step": 25812
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.574097135740971e-06,
+      "loss": 0.1358,
+      "step": 25813
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.561643835616438e-06,
+      "loss": 0.1675,
+      "step": 25814
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.549190535491905e-06,
+      "loss": 0.0736,
+      "step": 25815
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 3.536737235367372e-06,
+      "loss": 0.1161,
+      "step": 25816
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.5242839352428394e-06,
+      "loss": 0.1492,
+      "step": 25817
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.511830635118306e-06,
+      "loss": 0.0856,
+      "step": 25818
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.4993773349937734e-06,
+      "loss": 0.147,
+      "step": 25819
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.48692403486924e-06,
+      "loss": 0.1224,
+      "step": 25820
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.4744707347447074e-06,
+      "loss": 0.1071,
+      "step": 25821
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.462017434620174e-06,
+      "loss": 0.0515,
+      "step": 25822
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.449564134495641e-06,
+      "loss": 0.0716,
+      "step": 25823
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.437110834371108e-06,
+      "loss": 0.098,
+      "step": 25824
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.424657534246575e-06,
+      "loss": 0.146,
+      "step": 25825
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.412204234122042e-06,
+      "loss": 0.1761,
+      "step": 25826
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.399750933997509e-06,
+      "loss": 0.1125,
+      "step": 25827
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.387297633872976e-06,
+      "loss": 0.0853,
+      "step": 25828
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.374844333748443e-06,
+      "loss": 0.0998,
+      "step": 25829
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.36239103362391e-06,
+      "loss": 0.073,
+      "step": 25830
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.349937733499377e-06,
+      "loss": 0.2133,
+      "step": 25831
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.337484433374844e-06,
+      "loss": 0.0885,
+      "step": 25832
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.325031133250311e-06,
+      "loss": 0.1086,
+      "step": 25833
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.312577833125778e-06,
+      "loss": 0.0852,
+      "step": 25834
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.300124533001245e-06,
+      "loss": 0.2327,
+      "step": 25835
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.2876712328767123e-06,
+      "loss": 0.1019,
+      "step": 25836
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.275217932752179e-06,
+      "loss": 0.0893,
+      "step": 25837
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.2627646326276463e-06,
+      "loss": 0.1462,
+      "step": 25838
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.250311332503113e-06,
+      "loss": 0.1376,
+      "step": 25839
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.2378580323785803e-06,
+      "loss": 0.1118,
+      "step": 25840
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.225404732254047e-06,
+      "loss": 0.0637,
+      "step": 25841
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 3.2129514321295143e-06,
+      "loss": 0.0953,
+      "step": 25842
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.200498132004981e-06,
+      "loss": 0.1239,
+      "step": 25843
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.1880448318804483e-06,
+      "loss": 0.2374,
+      "step": 25844
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.175591531755915e-06,
+      "loss": 0.084,
+      "step": 25845
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.1631382316313823e-06,
+      "loss": 0.0877,
+      "step": 25846
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.150684931506849e-06,
+      "loss": 0.0727,
+      "step": 25847
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.138231631382316e-06,
+      "loss": 0.059,
+      "step": 25848
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.1257783312577827e-06,
+      "loss": 0.1177,
+      "step": 25849
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.11332503113325e-06,
+      "loss": 0.1614,
+      "step": 25850
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.1008717310087167e-06,
+      "loss": 0.0795,
+      "step": 25851
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.088418430884184e-06,
+      "loss": 0.0882,
+      "step": 25852
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.0759651307596507e-06,
+      "loss": 0.0719,
+      "step": 25853
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.063511830635118e-06,
+      "loss": 0.1726,
+      "step": 25854
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.051058530510585e-06,
+      "loss": 0.1142,
+      "step": 25855
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.038605230386052e-06,
+      "loss": 0.1055,
+      "step": 25856
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.026151930261519e-06,
+      "loss": 0.1523,
+      "step": 25857
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.013698630136986e-06,
+      "loss": 0.1066,
+      "step": 25858
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.001245330012453e-06,
+      "loss": 0.0834,
+      "step": 25859
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.98879202988792e-06,
+      "loss": 0.0874,
+      "step": 25860
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.976338729763387e-06,
+      "loss": 0.0443,
+      "step": 25861
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.963885429638854e-06,
+      "loss": 0.0715,
+      "step": 25862
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.951432129514321e-06,
+      "loss": 0.1558,
+      "step": 25863
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.938978829389788e-06,
+      "loss": 0.1554,
+      "step": 25864
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.926525529265255e-06,
+      "loss": 0.1192,
+      "step": 25865
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.914072229140722e-06,
+      "loss": 0.055,
+      "step": 25866
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.901618929016189e-06,
+      "loss": 0.1656,
+      "step": 25867
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.889165628891656e-06,
+      "loss": 0.0865,
+      "step": 25868
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.876712328767123e-06,
+      "loss": 0.1826,
+      "step": 25869
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.8642590286425904e-06,
+      "loss": 0.1569,
+      "step": 25870
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.851805728518057e-06,
+      "loss": 0.1479,
+      "step": 25871
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.8393524283935244e-06,
+      "loss": 0.082,
+      "step": 25872
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.826899128268991e-06,
+      "loss": 0.1006,
+      "step": 25873
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.8144458281444584e-06,
+      "loss": 0.062,
+      "step": 25874
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.8019925280199248e-06,
+      "loss": 0.1851,
+      "step": 25875
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.789539227895392e-06,
+      "loss": 0.235,
+      "step": 25876
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.7770859277708588e-06,
+      "loss": 0.1263,
+      "step": 25877
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.764632627646326e-06,
+      "loss": 0.0585,
+      "step": 25878
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.752179327521793e-06,
+      "loss": 0.0927,
+      "step": 25879
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.73972602739726e-06,
+      "loss": 0.0724,
+      "step": 25880
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.727272727272727e-06,
+      "loss": 0.0811,
+      "step": 25881
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.714819427148194e-06,
+      "loss": 0.1097,
+      "step": 25882
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.702366127023661e-06,
+      "loss": 0.0633,
+      "step": 25883
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.689912826899128e-06,
+      "loss": 0.0836,
+      "step": 25884
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.677459526774595e-06,
+      "loss": 0.0588,
+      "step": 25885
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.665006226650062e-06,
+      "loss": 0.1254,
+      "step": 25886
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.652552926525529e-06,
+      "loss": 0.0546,
+      "step": 25887
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.640099626400996e-06,
+      "loss": 0.129,
+      "step": 25888
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.6276463262764633e-06,
+      "loss": 0.1399,
+      "step": 25889
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.61519302615193e-06,
+      "loss": 0.1203,
+      "step": 25890
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.6027397260273973e-06,
+      "loss": 0.0802,
+      "step": 25891
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.590286425902864e-06,
+      "loss": 0.0998,
+      "step": 25892
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.5778331257783313e-06,
+      "loss": 0.0522,
+      "step": 25893
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 2.565379825653798e-06,
+      "loss": 0.1345,
+      "step": 25894
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.5529265255292653e-06,
+      "loss": 0.1316,
+      "step": 25895
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.540473225404732e-06,
+      "loss": 0.1292,
+      "step": 25896
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.5280199252801993e-06,
+      "loss": 0.1627,
+      "step": 25897
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.515566625155666e-06,
+      "loss": 0.0864,
+      "step": 25898
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.5031133250311333e-06,
+      "loss": 0.0695,
+      "step": 25899
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.4906600249066e-06,
+      "loss": 0.1027,
+      "step": 25900
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.4782067247820673e-06,
+      "loss": 0.1483,
+      "step": 25901
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.4657534246575337e-06,
+      "loss": 0.1428,
+      "step": 25902
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.453300124533001e-06,
+      "loss": 0.1061,
+      "step": 25903
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.4408468244084677e-06,
+      "loss": 0.114,
+      "step": 25904
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.428393524283935e-06,
+      "loss": 0.134,
+      "step": 25905
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.4159402241594017e-06,
+      "loss": 0.1132,
+      "step": 25906
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.403486924034869e-06,
+      "loss": 0.167,
+      "step": 25907
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.391033623910336e-06,
+      "loss": 0.1183,
+      "step": 25908
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.378580323785803e-06,
+      "loss": 0.0834,
+      "step": 25909
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.36612702366127e-06,
+      "loss": 0.1355,
+      "step": 25910
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.353673723536737e-06,
+      "loss": 0.0621,
+      "step": 25911
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.341220423412204e-06,
+      "loss": 0.0842,
+      "step": 25912
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.328767123287671e-06,
+      "loss": 0.179,
+      "step": 25913
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.316313823163138e-06,
+      "loss": 0.0967,
+      "step": 25914
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.303860523038605e-06,
+      "loss": 0.1086,
+      "step": 25915
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.291407222914072e-06,
+      "loss": 0.1409,
+      "step": 25916
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.278953922789539e-06,
+      "loss": 0.067,
+      "step": 25917
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.266500622665006e-06,
+      "loss": 0.0417,
+      "step": 25918
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.254047322540473e-06,
+      "loss": 0.1152,
+      "step": 25919
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.24159402241594e-06,
+      "loss": 0.1248,
+      "step": 25920
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.229140722291407e-06,
+      "loss": 0.144,
+      "step": 25921
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.2166874221668737e-06,
+      "loss": 0.1145,
+      "step": 25922
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.204234122042341e-06,
+      "loss": 0.0771,
+      "step": 25923
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.1917808219178078e-06,
+      "loss": 0.1444,
+      "step": 25924
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.179327521793275e-06,
+      "loss": 0.216,
+      "step": 25925
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.166874221668742e-06,
+      "loss": 0.2049,
+      "step": 25926
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.154420921544209e-06,
+      "loss": 0.1215,
+      "step": 25927
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.141967621419676e-06,
+      "loss": 0.1572,
+      "step": 25928
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.129514321295143e-06,
+      "loss": 0.0949,
+      "step": 25929
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.11706102117061e-06,
+      "loss": 0.1038,
+      "step": 25930
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.104607721046077e-06,
+      "loss": 0.0975,
+      "step": 25931
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.092154420921544e-06,
+      "loss": 0.0926,
+      "step": 25932
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.079701120797011e-06,
+      "loss": 0.0905,
+      "step": 25933
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.0672478206724782e-06,
+      "loss": 0.0979,
+      "step": 25934
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.054794520547945e-06,
+      "loss": 0.1136,
+      "step": 25935
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.042341220423412e-06,
+      "loss": 0.124,
+      "step": 25936
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.029887920298879e-06,
+      "loss": 0.1273,
+      "step": 25937
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.017434620174346e-06,
+      "loss": 0.2119,
+      "step": 25938
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.004981320049813e-06,
+      "loss": 0.148,
+      "step": 25939
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.99252801992528e-06,
+      "loss": 0.0656,
+      "step": 25940
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.980074719800747e-06,
+      "loss": 0.1283,
+      "step": 25941
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.9676214196762142e-06,
+      "loss": 0.1005,
+      "step": 25942
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.955168119551681e-06,
+      "loss": 0.1168,
+      "step": 25943
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.9427148194271483e-06,
+      "loss": 0.2471,
+      "step": 25944
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.930261519302615e-06,
+      "loss": 0.1337,
+      "step": 25945
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.9178082191780823e-06,
+      "loss": 0.0942,
+      "step": 25946
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.905354919053549e-06,
+      "loss": 0.1405,
+      "step": 25947
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.8929016189290158e-06,
+      "loss": 0.0692,
+      "step": 25948
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.8804483188044828e-06,
+      "loss": 0.1201,
+      "step": 25949
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.8679950186799499e-06,
+      "loss": 0.1236,
+      "step": 25950
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.8555417185554169e-06,
+      "loss": 0.2915,
+      "step": 25951
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.843088418430884e-06,
+      "loss": 0.0969,
+      "step": 25952
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.830635118306351e-06,
+      "loss": 0.1211,
+      "step": 25953
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.818181818181818e-06,
+      "loss": 0.0915,
+      "step": 25954
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.805728518057285e-06,
+      "loss": 0.0734,
+      "step": 25955
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.793275217932752e-06,
+      "loss": 0.0785,
+      "step": 25956
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.780821917808219e-06,
+      "loss": 0.1305,
+      "step": 25957
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.768368617683686e-06,
+      "loss": 0.107,
+      "step": 25958
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.755915317559153e-06,
+      "loss": 0.1436,
+      "step": 25959
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.74346201743462e-06,
+      "loss": 0.0694,
+      "step": 25960
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.731008717310087e-06,
+      "loss": 0.0923,
+      "step": 25961
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.718555417185554e-06,
+      "loss": 0.1358,
+      "step": 25962
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.706102117061021e-06,
+      "loss": 0.1415,
+      "step": 25963
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.693648816936488e-06,
+      "loss": 0.1336,
+      "step": 25964
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.681195516811955e-06,
+      "loss": 0.1436,
+      "step": 25965
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.668742216687422e-06,
+      "loss": 0.1281,
+      "step": 25966
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.656288916562889e-06,
+      "loss": 0.0933,
+      "step": 25967
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6438356164383561e-06,
+      "loss": 0.1366,
+      "step": 25968
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6313823163138231e-06,
+      "loss": 0.1408,
+      "step": 25969
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6189290161892901e-06,
+      "loss": 0.1036,
+      "step": 25970
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6064757160647571e-06,
+      "loss": 0.1045,
+      "step": 25971
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.5940224159402241e-06,
+      "loss": 0.1339,
+      "step": 25972
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.5815691158156912e-06,
+      "loss": 0.0925,
+      "step": 25973
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.569115815691158e-06,
+      "loss": 0.0953,
+      "step": 25974
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.556662515566625e-06,
+      "loss": 0.2169,
+      "step": 25975
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.544209215442092e-06,
+      "loss": 0.1157,
+      "step": 25976
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.531755915317559e-06,
+      "loss": 0.0678,
+      "step": 25977
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.519302615193026e-06,
+      "loss": 0.0595,
+      "step": 25978
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.506849315068493e-06,
+      "loss": 0.0857,
+      "step": 25979
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.49439601494396e-06,
+      "loss": 0.0756,
+      "step": 25980
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.481942714819427e-06,
+      "loss": 0.1365,
+      "step": 25981
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.469489414694894e-06,
+      "loss": 0.1339,
+      "step": 25982
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.457036114570361e-06,
+      "loss": 0.0705,
+      "step": 25983
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.444582814445828e-06,
+      "loss": 0.1022,
+      "step": 25984
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.4321295143212952e-06,
+      "loss": 0.0782,
+      "step": 25985
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.4196762141967622e-06,
+      "loss": 0.059,
+      "step": 25986
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.4072229140722292e-06,
+      "loss": 0.0954,
+      "step": 25987
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.394769613947696e-06,
+      "loss": 0.1433,
+      "step": 25988
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.382316313823163e-06,
+      "loss": 0.191,
+      "step": 25989
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.36986301369863e-06,
+      "loss": 0.0931,
+      "step": 25990
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.357409713574097e-06,
+      "loss": 0.2066,
+      "step": 25991
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.344956413449564e-06,
+      "loss": 0.1492,
+      "step": 25992
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.332503113325031e-06,
+      "loss": 0.0962,
+      "step": 25993
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.320049813200498e-06,
+      "loss": 0.1154,
+      "step": 25994
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.307596513075965e-06,
+      "loss": 0.1154,
+      "step": 25995
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.295143212951432e-06,
+      "loss": 0.0863,
+      "step": 25996
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.282689912826899e-06,
+      "loss": 0.0786,
+      "step": 25997
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.270236612702366e-06,
+      "loss": 0.14,
+      "step": 25998
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.257783312577833e-06,
+      "loss": 0.1004,
+      "step": 25999
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.2453300124533e-06,
+      "loss": 0.2514,
+      "step": 26000
+    },
+    {
+      "epoch": 9.97,
+      "eval_cer": 0.09989659948885031,
+      "eval_loss": 0.42421841621398926,
+      "eval_runtime": 47.3579,
+      "eval_samples_per_second": 37.016,
+      "eval_steps_per_second": 4.645,
+      "eval_wer": 0.15644657645834276,
+      "step": 26000
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.2328767123287668e-06,
+      "loss": 0.1163,
+      "step": 26001
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.2204234122042338e-06,
+      "loss": 0.1749,
+      "step": 26002
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.2079701120797008e-06,
+      "loss": 0.0969,
+      "step": 26003
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.195516811955168e-06,
+      "loss": 0.1262,
+      "step": 26004
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.183063511830635e-06,
+      "loss": 0.1519,
+      "step": 26005
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.170610211706102e-06,
+      "loss": 0.1716,
+      "step": 26006
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.158156911581569e-06,
+      "loss": 0.1212,
+      "step": 26007
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.145703611457036e-06,
+      "loss": 0.0977,
+      "step": 26008
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.133250311332503e-06,
+      "loss": 0.116,
+      "step": 26009
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.12079701120797e-06,
+      "loss": 0.1097,
+      "step": 26010
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.1083437110834369e-06,
+      "loss": 0.0728,
+      "step": 26011
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.0958904109589039e-06,
+      "loss": 0.0774,
+      "step": 26012
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.083437110834371e-06,
+      "loss": 0.2105,
+      "step": 26013
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.070983810709838e-06,
+      "loss": 0.0887,
+      "step": 26014
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.058530510585305e-06,
+      "loss": 0.1009,
+      "step": 26015
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.046077210460772e-06,
+      "loss": 0.117,
+      "step": 26016
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.0336239103362391e-06,
+      "loss": 0.06,
+      "step": 26017
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.021170610211706e-06,
+      "loss": 0.0759,
+      "step": 26018
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.008717310087173e-06,
+      "loss": 0.1701,
+      "step": 26019
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 9.9626400996264e-07,
+      "loss": 0.1191,
+      "step": 26020
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 9.838107098381071e-07,
+      "loss": 0.1114,
+      "step": 26021
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 9.713574097135741e-07,
+      "loss": 0.137,
+      "step": 26022
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 9.589041095890411e-07,
+      "loss": 0.0745,
+      "step": 26023
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 9.464508094645079e-07,
+      "loss": 0.1367,
+      "step": 26024
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 9.339975093399749e-07,
+      "loss": 0.1344,
+      "step": 26025
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 9.21544209215442e-07,
+      "loss": 0.104,
+      "step": 26026
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 9.09090909090909e-07,
+      "loss": 0.0993,
+      "step": 26027
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.96637608966376e-07,
+      "loss": 0.0977,
+      "step": 26028
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.84184308841843e-07,
+      "loss": 0.0801,
+      "step": 26029
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.7173100871731e-07,
+      "loss": 0.1689,
+      "step": 26030
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.59277708592777e-07,
+      "loss": 0.0904,
+      "step": 26031
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.46824408468244e-07,
+      "loss": 0.1686,
+      "step": 26032
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.34371108343711e-07,
+      "loss": 0.1113,
+      "step": 26033
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.219178082191781e-07,
+      "loss": 0.0894,
+      "step": 26034
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 8.094645080946451e-07,
+      "loss": 0.203,
+      "step": 26035
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.970112079701121e-07,
+      "loss": 0.0784,
+      "step": 26036
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.84557907845579e-07,
+      "loss": 0.0734,
+      "step": 26037
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.72104607721046e-07,
+      "loss": 0.1877,
+      "step": 26038
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.59651307596513e-07,
+      "loss": 0.147,
+      "step": 26039
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.4719800747198e-07,
+      "loss": 0.0691,
+      "step": 26040
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.34744707347447e-07,
+      "loss": 0.072,
+      "step": 26041
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.22291407222914e-07,
+      "loss": 0.1485,
+      "step": 26042
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 7.098381070983811e-07,
+      "loss": 0.0686,
+      "step": 26043
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.97384806973848e-07,
+      "loss": 0.178,
+      "step": 26044
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.84931506849315e-07,
+      "loss": 0.1684,
+      "step": 26045
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.72478206724782e-07,
+      "loss": 0.102,
+      "step": 26046
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.60024906600249e-07,
+      "loss": 0.0629,
+      "step": 26047
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.47571606475716e-07,
+      "loss": 0.1261,
+      "step": 26048
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.35118306351183e-07,
+      "loss": 0.079,
+      "step": 26049
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 6.2266500622665e-07,
+      "loss": 0.2342,
+      "step": 26050
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 6.102117061021169e-07,
+      "loss": 0.2353,
+      "step": 26051
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.97758405977584e-07,
+      "loss": 0.128,
+      "step": 26052
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.85305105853051e-07,
+      "loss": 0.1057,
+      "step": 26053
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.72851805728518e-07,
+      "loss": 0.102,
+      "step": 26054
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.60398505603985e-07,
+      "loss": 0.0705,
+      "step": 26055
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.479452054794519e-07,
+      "loss": 0.0935,
+      "step": 26056
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.35491905354919e-07,
+      "loss": 0.2035,
+      "step": 26057
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.23038605230386e-07,
+      "loss": 0.0792,
+      "step": 26058
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 5.10585305105853e-07,
+      "loss": 0.1307,
+      "step": 26059
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.9813200498132e-07,
+      "loss": 0.0897,
+      "step": 26060
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.856787048567871e-07,
+      "loss": 0.091,
+      "step": 26061
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.7322540473225396e-07,
+      "loss": 0.1151,
+      "step": 26062
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.60772104607721e-07,
+      "loss": 0.2117,
+      "step": 26063
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.48318804483188e-07,
+      "loss": 0.202,
+      "step": 26064
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.35865504358655e-07,
+      "loss": 0.0756,
+      "step": 26065
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.23412204234122e-07,
+      "loss": 0.0756,
+      "step": 26066
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 4.1095890410958903e-07,
+      "loss": 0.0725,
+      "step": 26067
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.9850560398505604e-07,
+      "loss": 0.0747,
+      "step": 26068
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.86052303860523e-07,
+      "loss": 0.1711,
+      "step": 26069
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.7359900373599e-07,
+      "loss": 0.1213,
+      "step": 26070
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.61145703611457e-07,
+      "loss": 0.1042,
+      "step": 26071
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.48692403486924e-07,
+      "loss": 0.0698,
+      "step": 26072
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.36239103362391e-07,
+      "loss": 0.1369,
+      "step": 26073
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.23785803237858e-07,
+      "loss": 0.1745,
+      "step": 26074
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 3.11332503113325e-07,
+      "loss": 0.1339,
+      "step": 26075
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 2.98879202988792e-07,
+      "loss": 0.1597,
+      "step": 26076
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.86425902864259e-07,
+      "loss": 0.1088,
+      "step": 26077
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.7397260273972597e-07,
+      "loss": 0.1115,
+      "step": 26078
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.61519302615193e-07,
+      "loss": 0.0942,
+      "step": 26079
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.4906600249066e-07,
+      "loss": 0.0612,
+      "step": 26080
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.3661270236612698e-07,
+      "loss": 0.1135,
+      "step": 26081
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.24159402241594e-07,
+      "loss": 0.1591,
+      "step": 26082
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.11706102117061e-07,
+      "loss": 0.0977,
+      "step": 26083
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.9925280199252802e-07,
+      "loss": 0.1231,
+      "step": 26084
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.86799501867995e-07,
+      "loss": 0.118,
+      "step": 26085
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.74346201743462e-07,
+      "loss": 0.0891,
+      "step": 26086
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.61892901618929e-07,
+      "loss": 0.2401,
+      "step": 26087
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.49439601494396e-07,
+      "loss": 0.1349,
+      "step": 26088
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.3698630136986298e-07,
+      "loss": 0.1278,
+      "step": 26089
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.2453300124533e-07,
+      "loss": 0.1628,
+      "step": 26090
+    },
+    {
+      "epoch": 10.0,
+      "step": 26090,
+      "total_flos": 5.670426185293963e+20,
+      "train_loss": 0.30801558772101284,
+      "train_runtime": 38596.418,
+      "train_samples_per_second": 43.262,
+      "train_steps_per_second": 0.676
+    }
+  ],
+  "max_steps": 26090,
+  "num_train_epochs": 10,
+  "total_flos": 5.670426185293963e+20,
+  "trial_name": null,
+  "trial_params": null
+}