diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,32227 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 5352,
+  "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": 5.6648,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.676,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-06,
+      "loss": 5.9349,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-06,
+      "loss": 5.3714,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-06,
+      "loss": 5.6428,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 5.4509,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-06,
+      "loss": 5.4726,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 5.2373,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2e-05,
+      "loss": 5.4474,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4e-05,
+      "loss": 5.0202,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6e-05,
+      "loss": 4.9822,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 4.9562,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 5.0026,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2e-05,
+      "loss": 4.6385,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 4.7154,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6e-05,
+      "loss": 4.7762,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8e-05,
+      "loss": 4.8101,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 4.6273,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2e-05,
+      "loss": 4.5567,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 4.8683,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-05,
+      "loss": 4.5065,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8e-05,
+      "loss": 4.9547,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 4.6395,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2000000000000004e-05,
+      "loss": 4.4258,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4e-05,
+      "loss": 4.5337,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6e-05,
+      "loss": 4.6985,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8e-05,
+      "loss": 4.4575,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 4.1981,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2e-05,
+      "loss": 4.5233,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4e-05,
+      "loss": 4.495,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6e-05,
+      "loss": 4.4177,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.800000000000001e-05,
+      "loss": 4.7941,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-05,
+      "loss": 4.4348,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2e-05,
+      "loss": 4.2586,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4e-05,
+      "loss": 4.5347,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.6e-05,
+      "loss": 4.7708,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 4.4374,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.000000000000001e-05,
+      "loss": 4.5724,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2e-05,
+      "loss": 4.627,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4e-05,
+      "loss": 4.6501,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.6e-05,
+      "loss": 4.4484,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.8e-05,
+      "loss": 4.3345,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8e-05,
+      "loss": 4.569,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.2e-05,
+      "loss": 4.3274,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.400000000000001e-05,
+      "loss": 4.739,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.599999999999999e-05,
+      "loss": 4.654,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.8e-05,
+      "loss": 4.7521,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 4.8937,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.2e-05,
+      "loss": 4.9379,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.400000000000001e-05,
+      "loss": 5.0091,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.6e-05,
+      "loss": 4.4864,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.800000000000001e-05,
+      "loss": 4.6497,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 4.5328,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000102,
+      "loss": 4.4313,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000104,
+      "loss": 4.5186,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000106,
+      "loss": 4.555,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000108,
+      "loss": 4.5202,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011,
+      "loss": 4.4346,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000112,
+      "loss": 4.4787,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000114,
+      "loss": 4.4381,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00011600000000000001,
+      "loss": 4.5978,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000118,
+      "loss": 4.5152,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00012,
+      "loss": 4.8357,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000122,
+      "loss": 4.2671,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000124,
+      "loss": 4.503,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000126,
+      "loss": 4.6185,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000128,
+      "loss": 4.5785,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 4.5626,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000132,
+      "loss": 4.5544,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000134,
+      "loss": 4.4384,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 4.5741,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00013800000000000002,
+      "loss": 4.2417,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00014000000000000001,
+      "loss": 4.6551,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00014199999999999998,
+      "loss": 4.3695,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000144,
+      "loss": 4.8736,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000146,
+      "loss": 4.6611,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000148,
+      "loss": 4.4942,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00015,
+      "loss": 4.4072,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000152,
+      "loss": 4.4638,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000154,
+      "loss": 4.6975,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000156,
+      "loss": 4.4601,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000158,
+      "loss": 4.6589,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016,
+      "loss": 4.3927,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000162,
+      "loss": 4.6753,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000164,
+      "loss": 4.5663,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016600000000000002,
+      "loss": 4.4293,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00016800000000000002,
+      "loss": 4.7984,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017,
+      "loss": 4.4343,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017199999999999998,
+      "loss": 4.7634,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000174,
+      "loss": 4.3463,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000176,
+      "loss": 4.7658,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000178,
+      "loss": 4.4552,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 4.296,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000182,
+      "loss": 4.7972,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000184,
+      "loss": 4.9151,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000186,
+      "loss": 4.8277,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00018800000000000002,
+      "loss": 5.0903,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019,
+      "loss": 4.6821,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000192,
+      "loss": 4.9578,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000194,
+      "loss": 5.0353,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019600000000000002,
+      "loss": 4.814,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 4.7056,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 4.929,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000202,
+      "loss": 4.7043,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000204,
+      "loss": 4.7845,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000206,
+      "loss": 4.6563,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000208,
+      "loss": 4.7215,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00021,
+      "loss": 4.7869,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000212,
+      "loss": 4.7671,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000214,
+      "loss": 4.878,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000216,
+      "loss": 5.169,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000218,
+      "loss": 4.6003,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00022,
+      "loss": 4.8139,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000222,
+      "loss": 4.5597,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000224,
+      "loss": 4.4441,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00022600000000000002,
+      "loss": 4.6085,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000228,
+      "loss": 4.9228,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00023,
+      "loss": 5.0918,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00023200000000000003,
+      "loss": 4.8138,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00023400000000000002,
+      "loss": 5.094,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000236,
+      "loss": 5.0022,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00023799999999999998,
+      "loss": 4.8353,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00024,
+      "loss": 4.7763,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000242,
+      "loss": 5.0577,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000244,
+      "loss": 4.6117,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000246,
+      "loss": 5.0354,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000248,
+      "loss": 4.7871,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00025,
+      "loss": 5.0224,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000252,
+      "loss": 5.0258,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000254,
+      "loss": 5.9592,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000256,
+      "loss": 6.0343,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00025800000000000004,
+      "loss": 4.718,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 6.0153,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000262,
+      "loss": 5.1783,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000264,
+      "loss": 4.6485,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000266,
+      "loss": 5.1805,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000268,
+      "loss": 5.1717,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00027,
+      "loss": 5.1344,
+      "step": 138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 6.2138,
+      "step": 139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00027400000000000005,
+      "loss": 4.8561,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00027600000000000004,
+      "loss": 5.5285,
+      "step": 141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00027800000000000004,
+      "loss": 5.22,
+      "step": 142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00028000000000000003,
+      "loss": 4.9118,
+      "step": 143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 6.0522,
+      "step": 144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00028399999999999996,
+      "loss": 5.4507,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00028599999999999996,
+      "loss": 5.1606,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000288,
+      "loss": 5.2633,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029,
+      "loss": 5.6957,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000292,
+      "loss": 5.8686,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000294,
+      "loss": 7.9547,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000296,
+      "loss": 5.163,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000298,
+      "loss": 5.6928,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0003,
+      "loss": 4.7312,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000302,
+      "loss": 6.1422,
+      "step": 154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000304,
+      "loss": 5.0531,
+      "step": 155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000306,
+      "loss": 5.2937,
+      "step": 156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000308,
+      "loss": 5.1818,
+      "step": 157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00031,
+      "loss": 5.0077,
+      "step": 158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000312,
+      "loss": 5.3497,
+      "step": 159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000314,
+      "loss": 6.3285,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000316,
+      "loss": 5.3665,
+      "step": 161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00031800000000000003,
+      "loss": 5.377,
+      "step": 162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00032,
+      "loss": 5.1629,
+      "step": 163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000322,
+      "loss": 5.3016,
+      "step": 164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000324,
+      "loss": 5.1226,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000326,
+      "loss": 5.3764,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000328,
+      "loss": 4.9585,
+      "step": 167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00033,
+      "loss": 4.8569,
+      "step": 168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00033200000000000005,
+      "loss": 4.7924,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00033400000000000004,
+      "loss": 4.9276,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00033600000000000004,
+      "loss": 5.3661,
+      "step": 171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00033800000000000003,
+      "loss": 6.4392,
+      "step": 172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00034,
+      "loss": 4.9998,
+      "step": 173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000342,
+      "loss": 5.1803,
+      "step": 174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00034399999999999996,
+      "loss": 5.7304,
+      "step": 175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000346,
+      "loss": 5.1821,
+      "step": 176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000348,
+      "loss": 5.0313,
+      "step": 177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00035,
+      "loss": 4.8978,
+      "step": 178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000352,
+      "loss": 4.7998,
+      "step": 179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000354,
+      "loss": 5.6087,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000356,
+      "loss": 5.3862,
+      "step": 181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000358,
+      "loss": 5.4141,
+      "step": 182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 6.2339,
+      "step": 183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000362,
+      "loss": 5.1922,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000364,
+      "loss": 5.2494,
+      "step": 185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000366,
+      "loss": 5.5893,
+      "step": 186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000368,
+      "loss": 4.9388,
+      "step": 187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00037,
+      "loss": 5.3492,
+      "step": 188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000372,
+      "loss": 4.9751,
+      "step": 189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000374,
+      "loss": 6.166,
+      "step": 190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00037600000000000003,
+      "loss": 6.3089,
+      "step": 191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000378,
+      "loss": 6.0773,
+      "step": 192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00038,
+      "loss": 6.5369,
+      "step": 193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000382,
+      "loss": 5.1475,
+      "step": 194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000384,
+      "loss": 5.3076,
+      "step": 195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000386,
+      "loss": 6.845,
+      "step": 196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000388,
+      "loss": 5.578,
+      "step": 197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 4.9703,
+      "step": 198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 6.9593,
+      "step": 199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039200000000000004,
+      "loss": 7.1339,
+      "step": 200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039400000000000004,
+      "loss": 6.2173,
+      "step": 201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 5.1485,
+      "step": 202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000398,
+      "loss": 5.4666,
+      "step": 203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0004,
+      "loss": 5.9378,
+      "step": 204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000402,
+      "loss": 5.0954,
+      "step": 205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000404,
+      "loss": 6.1447,
+      "step": 206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00040600000000000006,
+      "loss": 5.2555,
+      "step": 207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000408,
+      "loss": 5.9748,
+      "step": 208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00041,
+      "loss": 5.7742,
+      "step": 209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000412,
+      "loss": 5.5892,
+      "step": 210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000414,
+      "loss": 5.2607,
+      "step": 211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000416,
+      "loss": 5.6333,
+      "step": 212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00041799999999999997,
+      "loss": 5.8993,
+      "step": 213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00042,
+      "loss": 4.9825,
+      "step": 214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000422,
+      "loss": 6.4526,
+      "step": 215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000424,
+      "loss": 4.8881,
+      "step": 216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000426,
+      "loss": 5.6929,
+      "step": 217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000428,
+      "loss": 4.9957,
+      "step": 218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00043,
+      "loss": 5.8813,
+      "step": 219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000432,
+      "loss": 5.9127,
+      "step": 220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00043400000000000003,
+      "loss": 5.4142,
+      "step": 221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000436,
+      "loss": 6.0848,
+      "step": 222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000438,
+      "loss": 5.0846,
+      "step": 223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00044,
+      "loss": 5.2926,
+      "step": 224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000442,
+      "loss": 5.45,
+      "step": 225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000444,
+      "loss": 6.0783,
+      "step": 226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000446,
+      "loss": 5.4007,
+      "step": 227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000448,
+      "loss": 5.4221,
+      "step": 228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 5.2132,
+      "step": 229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00045200000000000004,
+      "loss": 5.1292,
+      "step": 230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00045400000000000003,
+      "loss": 5.2217,
+      "step": 231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000456,
+      "loss": 7.1036,
+      "step": 232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000458,
+      "loss": 5.8019,
+      "step": 233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00046,
+      "loss": 5.8073,
+      "step": 234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000462,
+      "loss": 4.9821,
+      "step": 235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00046400000000000006,
+      "loss": 5.9624,
+      "step": 236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00046600000000000005,
+      "loss": 5.2112,
+      "step": 237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00046800000000000005,
+      "loss": 6.3198,
+      "step": 238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00047,
+      "loss": 6.1792,
+      "step": 239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000472,
+      "loss": 6.9,
+      "step": 240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000474,
+      "loss": 5.3152,
+      "step": 241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047599999999999997,
+      "loss": 5.9639,
+      "step": 242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00047799999999999996,
+      "loss": 7.2942,
+      "step": 243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00048,
+      "loss": 6.2327,
+      "step": 244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000482,
+      "loss": 5.3996,
+      "step": 245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000484,
+      "loss": 5.5747,
+      "step": 246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000486,
+      "loss": 5.8373,
+      "step": 247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000488,
+      "loss": 5.6701,
+      "step": 248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00049,
+      "loss": 5.6463,
+      "step": 249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000492,
+      "loss": 7.3198,
+      "step": 250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000494,
+      "loss": 5.5688,
+      "step": 251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000496,
+      "loss": 6.4274,
+      "step": 252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000498,
+      "loss": 6.069,
+      "step": 253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005,
+      "loss": 5.464,
+      "step": 254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0005020000000000001,
+      "loss": 5.7453,
+      "step": 255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000504,
+      "loss": 5.6692,
+      "step": 256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000506,
+      "loss": 5.5643,
+      "step": 257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000508,
+      "loss": 6.9768,
+      "step": 258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00051,
+      "loss": 5.6372,
+      "step": 259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000512,
+      "loss": 4.9631,
+      "step": 260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000514,
+      "loss": 5.9464,
+      "step": 261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005160000000000001,
+      "loss": 7.3611,
+      "step": 262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000518,
+      "loss": 5.7614,
+      "step": 263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005200000000000001,
+      "loss": 6.101,
+      "step": 264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000522,
+      "loss": 5.6324,
+      "step": 265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000524,
+      "loss": 5.1006,
+      "step": 266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000526,
+      "loss": 6.4633,
+      "step": 267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000528,
+      "loss": 5.757,
+      "step": 268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005300000000000001,
+      "loss": 5.7442,
+      "step": 269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000532,
+      "loss": 5.1216,
+      "step": 270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005340000000000001,
+      "loss": 5.345,
+      "step": 271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000536,
+      "loss": 5.3651,
+      "step": 272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005380000000000001,
+      "loss": 5.7791,
+      "step": 273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00054,
+      "loss": 6.4297,
+      "step": 274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005420000000000001,
+      "loss": 6.2293,
+      "step": 275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0005440000000000001,
+      "loss": 6.3551,
+      "step": 276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000546,
+      "loss": 7.282,
+      "step": 277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005480000000000001,
+      "loss": 6.244,
+      "step": 278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00055,
+      "loss": 5.2515,
+      "step": 279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005520000000000001,
+      "loss": 5.2625,
+      "step": 280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000554,
+      "loss": 6.3458,
+      "step": 281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005560000000000001,
+      "loss": 6.6737,
+      "step": 282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000558,
+      "loss": 6.0929,
+      "step": 283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005600000000000001,
+      "loss": 5.5944,
+      "step": 284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005620000000000001,
+      "loss": 5.738,
+      "step": 285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005639999999999999,
+      "loss": 5.45,
+      "step": 286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000566,
+      "loss": 5.834,
+      "step": 287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005679999999999999,
+      "loss": 5.9889,
+      "step": 288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00057,
+      "loss": 5.1873,
+      "step": 289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0005719999999999999,
+      "loss": 6.4616,
+      "step": 290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000574,
+      "loss": 5.6644,
+      "step": 291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000576,
+      "loss": 7.4519,
+      "step": 292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000578,
+      "loss": 5.5747,
+      "step": 293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00058,
+      "loss": 5.6382,
+      "step": 294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005819999999999999,
+      "loss": 5.4684,
+      "step": 295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000584,
+      "loss": 8.1853,
+      "step": 296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0005859999999999999,
+      "loss": 5.8813,
+      "step": 297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000588,
+      "loss": 5.5969,
+      "step": 298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00059,
+      "loss": 5.3631,
+      "step": 299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000592,
+      "loss": 7.4291,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000594,
+      "loss": 6.1705,
+      "step": 301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000596,
+      "loss": 5.969,
+      "step": 302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000598,
+      "loss": 5.6395,
+      "step": 303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0006,
+      "loss": 5.4777,
+      "step": 304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000602,
+      "loss": 6.5724,
+      "step": 305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000604,
+      "loss": 5.983,
+      "step": 306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000606,
+      "loss": 5.5326,
+      "step": 307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000608,
+      "loss": 8.072,
+      "step": 308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00061,
+      "loss": 6.0716,
+      "step": 309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000612,
+      "loss": 6.3669,
+      "step": 310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000614,
+      "loss": 5.6016,
+      "step": 311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000616,
+      "loss": 5.8001,
+      "step": 312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006180000000000001,
+      "loss": 5.7067,
+      "step": 313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00062,
+      "loss": 6.1678,
+      "step": 314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000622,
+      "loss": 5.4317,
+      "step": 315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000624,
+      "loss": 5.3798,
+      "step": 316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000626,
+      "loss": 5.4522,
+      "step": 317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000628,
+      "loss": 5.3276,
+      "step": 318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00063,
+      "loss": 6.4394,
+      "step": 319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000632,
+      "loss": 6.2204,
+      "step": 320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000634,
+      "loss": 6.4072,
+      "step": 321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006360000000000001,
+      "loss": 10.5052,
+      "step": 322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000638,
+      "loss": 6.3739,
+      "step": 323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00064,
+      "loss": 5.9886,
+      "step": 324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000642,
+      "loss": 5.6845,
+      "step": 325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000644,
+      "loss": 5.8231,
+      "step": 326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000646,
+      "loss": 5.8659,
+      "step": 327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000648,
+      "loss": 5.9409,
+      "step": 328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0006500000000000001,
+      "loss": 5.4434,
+      "step": 329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000652,
+      "loss": 5.9028,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006540000000000001,
+      "loss": 5.4822,
+      "step": 331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000656,
+      "loss": 5.7519,
+      "step": 332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006580000000000001,
+      "loss": 7.2035,
+      "step": 333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00066,
+      "loss": 6.2942,
+      "step": 334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000662,
+      "loss": 5.8861,
+      "step": 335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006640000000000001,
+      "loss": 6.1595,
+      "step": 336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000666,
+      "loss": 5.6132,
+      "step": 337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006680000000000001,
+      "loss": 5.5317,
+      "step": 338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00067,
+      "loss": 5.3997,
+      "step": 339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006720000000000001,
+      "loss": 5.6456,
+      "step": 340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000674,
+      "loss": 5.3102,
+      "step": 341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006760000000000001,
+      "loss": 5.799,
+      "step": 342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006780000000000001,
+      "loss": 5.7972,
+      "step": 343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00068,
+      "loss": 5.2866,
+      "step": 344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006820000000000001,
+      "loss": 6.5827,
+      "step": 345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000684,
+      "loss": 5.4417,
+      "step": 346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006860000000000001,
+      "loss": 6.4295,
+      "step": 347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006879999999999999,
+      "loss": 5.8489,
+      "step": 348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00069,
+      "loss": 8.8265,
+      "step": 349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000692,
+      "loss": 7.9957,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000694,
+      "loss": 7.0049,
+      "step": 351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000696,
+      "loss": 5.9574,
+      "step": 352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006979999999999999,
+      "loss": 5.69,
+      "step": 353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007,
+      "loss": 5.813,
+      "step": 354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007019999999999999,
+      "loss": 5.601,
+      "step": 355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000704,
+      "loss": 6.706,
+      "step": 356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007059999999999999,
+      "loss": 5.6837,
+      "step": 357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000708,
+      "loss": 6.5152,
+      "step": 358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00071,
+      "loss": 6.9277,
+      "step": 359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000712,
+      "loss": 5.7925,
+      "step": 360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000714,
+      "loss": 6.1777,
+      "step": 361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000716,
+      "loss": 5.3884,
+      "step": 362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000718,
+      "loss": 6.0113,
+      "step": 363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0007199999999999999,
+      "loss": 6.076,
+      "step": 364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000722,
+      "loss": 7.313,
+      "step": 365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000724,
+      "loss": 6.3203,
+      "step": 366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000726,
+      "loss": 5.5156,
+      "step": 367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000728,
+      "loss": 5.7185,
+      "step": 368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00073,
+      "loss": 6.8319,
+      "step": 369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000732,
+      "loss": 6.0078,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000734,
+      "loss": 5.5134,
+      "step": 371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000736,
+      "loss": 6.1058,
+      "step": 372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000738,
+      "loss": 5.8369,
+      "step": 373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00074,
+      "loss": 6.0533,
+      "step": 374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000742,
+      "loss": 5.3299,
+      "step": 375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000744,
+      "loss": 5.7861,
+      "step": 376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000746,
+      "loss": 5.5835,
+      "step": 377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000748,
+      "loss": 5.4525,
+      "step": 378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00075,
+      "loss": 5.543,
+      "step": 379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0007520000000000001,
+      "loss": 6.227,
+      "step": 380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000754,
+      "loss": 5.4066,
+      "step": 381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000756,
+      "loss": 5.5899,
+      "step": 382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000758,
+      "loss": 5.6576,
+      "step": 383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00076,
+      "loss": 6.3899,
+      "step": 384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000762,
+      "loss": 5.7219,
+      "step": 385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000764,
+      "loss": 5.7672,
+      "step": 386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007660000000000001,
+      "loss": 7.8557,
+      "step": 387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000768,
+      "loss": 5.1948,
+      "step": 388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007700000000000001,
+      "loss": 7.3807,
+      "step": 389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000772,
+      "loss": 7.8927,
+      "step": 390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007740000000000001,
+      "loss": 5.3207,
+      "step": 391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000776,
+      "loss": 6.2152,
+      "step": 392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000778,
+      "loss": 5.6751,
+      "step": 393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007800000000000001,
+      "loss": 6.2671,
+      "step": 394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000782,
+      "loss": 5.7835,
+      "step": 395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007840000000000001,
+      "loss": 6.0475,
+      "step": 396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000786,
+      "loss": 7.9274,
+      "step": 397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007880000000000001,
+      "loss": 10.0216,
+      "step": 398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00079,
+      "loss": 13.0074,
+      "step": 399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007920000000000001,
+      "loss": 6.1108,
+      "step": 400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0007940000000000001,
+      "loss": 6.0387,
+      "step": 401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000796,
+      "loss": 6.0469,
+      "step": 402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0007980000000000001,
+      "loss": 5.5738,
+      "step": 403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008,
+      "loss": 5.8432,
+      "step": 404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008020000000000001,
+      "loss": 5.5828,
+      "step": 405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000804,
+      "loss": 6.5185,
+      "step": 406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008060000000000001,
+      "loss": 5.9822,
+      "step": 407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000808,
+      "loss": 5.4748,
+      "step": 408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008100000000000001,
+      "loss": 6.1908,
+      "step": 409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008120000000000001,
+      "loss": 6.0397,
+      "step": 410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008139999999999999,
+      "loss": 6.2511,
+      "step": 411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000816,
+      "loss": 6.0171,
+      "step": 412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008179999999999999,
+      "loss": 6.219,
+      "step": 413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00082,
+      "loss": 6.3389,
+      "step": 414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008219999999999999,
+      "loss": 6.3234,
+      "step": 415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000824,
+      "loss": 5.8175,
+      "step": 416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000826,
+      "loss": 6.3443,
+      "step": 417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000828,
+      "loss": 5.8092,
+      "step": 418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00083,
+      "loss": 5.9956,
+      "step": 419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000832,
+      "loss": 6.0077,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000834,
+      "loss": 5.6281,
+      "step": 421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008359999999999999,
+      "loss": 5.7333,
+      "step": 422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000838,
+      "loss": 6.1026,
+      "step": 423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00084,
+      "loss": 6.0927,
+      "step": 424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000842,
+      "loss": 5.6681,
+      "step": 425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000844,
+      "loss": 5.6647,
+      "step": 426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000846,
+      "loss": 7.0708,
+      "step": 427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000848,
+      "loss": 8.3091,
+      "step": 428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00085,
+      "loss": 6.2796,
+      "step": 429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000852,
+      "loss": 5.7657,
+      "step": 430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000854,
+      "loss": 6.2377,
+      "step": 431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000856,
+      "loss": 6.0797,
+      "step": 432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000858,
+      "loss": 5.745,
+      "step": 433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00086,
+      "loss": 5.7692,
+      "step": 434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000862,
+      "loss": 5.6951,
+      "step": 435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000864,
+      "loss": 5.9647,
+      "step": 436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000866,
+      "loss": 5.825,
+      "step": 437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008680000000000001,
+      "loss": 6.3786,
+      "step": 438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00087,
+      "loss": 5.4627,
+      "step": 439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000872,
+      "loss": 5.7975,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000874,
+      "loss": 5.7407,
+      "step": 441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000876,
+      "loss": 6.7446,
+      "step": 442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000878,
+      "loss": 5.4677,
+      "step": 443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00088,
+      "loss": 5.9567,
+      "step": 444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000882,
+      "loss": 5.971,
+      "step": 445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000884,
+      "loss": 9.049,
+      "step": 446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008860000000000001,
+      "loss": 7.7203,
+      "step": 447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000888,
+      "loss": 6.2808,
+      "step": 448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008900000000000001,
+      "loss": 8.3944,
+      "step": 449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000892,
+      "loss": 5.6316,
+      "step": 450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000894,
+      "loss": 5.674,
+      "step": 451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000896,
+      "loss": 6.1824,
+      "step": 452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000898,
+      "loss": 7.5139,
+      "step": 453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0009000000000000001,
+      "loss": 5.7471,
+      "step": 454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000902,
+      "loss": 6.7631,
+      "step": 455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009040000000000001,
+      "loss": 5.7631,
+      "step": 456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000906,
+      "loss": 5.4331,
+      "step": 457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009080000000000001,
+      "loss": 6.6645,
+      "step": 458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00091,
+      "loss": 5.651,
+      "step": 459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000912,
+      "loss": 6.4654,
+      "step": 460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009140000000000001,
+      "loss": 6.6664,
+      "step": 461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000916,
+      "loss": 6.019,
+      "step": 462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009180000000000001,
+      "loss": 6.3514,
+      "step": 463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00092,
+      "loss": 5.9228,
+      "step": 464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009220000000000001,
+      "loss": 5.5557,
+      "step": 465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000924,
+      "loss": 6.1756,
+      "step": 466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009260000000000001,
+      "loss": 5.8819,
+      "step": 467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009280000000000001,
+      "loss": 6.0711,
+      "step": 468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00093,
+      "loss": 6.7006,
+      "step": 469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009320000000000001,
+      "loss": 6.3118,
+      "step": 470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000934,
+      "loss": 6.613,
+      "step": 471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0009360000000000001,
+      "loss": 5.9123,
+      "step": 472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 5.569,
+      "step": 473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00094,
+      "loss": 6.9474,
+      "step": 474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000942,
+      "loss": 6.2632,
+      "step": 475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000944,
+      "loss": 7.1214,
+      "step": 476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000946,
+      "loss": 6.7688,
+      "step": 477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000948,
+      "loss": 5.7178,
+      "step": 478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00095,
+      "loss": 7.2315,
+      "step": 479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009519999999999999,
+      "loss": 9.2455,
+      "step": 480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000954,
+      "loss": 6.3192,
+      "step": 481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009559999999999999,
+      "loss": 7.9877,
+      "step": 482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000958,
+      "loss": 6.3999,
+      "step": 483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00096,
+      "loss": 6.0317,
+      "step": 484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000962,
+      "loss": 6.7538,
+      "step": 485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000964,
+      "loss": 7.5668,
+      "step": 486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000966,
+      "loss": 5.9268,
+      "step": 487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000968,
+      "loss": 6.243,
+      "step": 488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0009699999999999999,
+      "loss": 6.1745,
+      "step": 489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000972,
+      "loss": 5.8666,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000974,
+      "loss": 5.6379,
+      "step": 491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000976,
+      "loss": 7.3065,
+      "step": 492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000978,
+      "loss": 6.6461,
+      "step": 493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00098,
+      "loss": 8.1153,
+      "step": 494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000982,
+      "loss": 7.2819,
+      "step": 495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000984,
+      "loss": 6.0571,
+      "step": 496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009860000000000001,
+      "loss": 5.6583,
+      "step": 497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000988,
+      "loss": 8.7847,
+      "step": 498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00099,
+      "loss": 5.8858,
+      "step": 499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000992,
+      "loss": 5.6407,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 6.388172149658203,
+      "eval_runtime": 1251.8889,
+      "eval_samples_per_second": 2.11,
+      "eval_steps_per_second": 0.264,
+      "eval_wer": 1.9939309797699325,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000994,
+      "loss": 6.3821,
+      "step": 501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000996,
+      "loss": 6.0271,
+      "step": 502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000998,
+      "loss": 6.2827,
+      "step": 503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.001,
+      "loss": 6.0016,
+      "step": 504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009997938994229184,
+      "loss": 6.7652,
+      "step": 505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009995877988458368,
+      "loss": 6.0915,
+      "step": 506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009993816982687552,
+      "loss": 5.8676,
+      "step": 507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0009991755976916736,
+      "loss": 5.9827,
+      "step": 508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000998969497114592,
+      "loss": 6.399,
+      "step": 509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009987633965375105,
+      "loss": 6.1642,
+      "step": 510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009985572959604287,
+      "loss": 5.8647,
+      "step": 511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000998351195383347,
+      "loss": 6.3521,
+      "step": 512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009981450948062655,
+      "loss": 9.7476,
+      "step": 513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009979389942291839,
+      "loss": 5.5993,
+      "step": 514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009977328936521023,
+      "loss": 5.8146,
+      "step": 515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009975267930750207,
+      "loss": 7.2953,
+      "step": 516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000997320692497939,
+      "loss": 6.1749,
+      "step": 517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009971145919208575,
+      "loss": 6.2477,
+      "step": 518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000996908491343776,
+      "loss": 6.3204,
+      "step": 519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009967023907666943,
+      "loss": 6.0038,
+      "step": 520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009964962901896125,
+      "loss": 5.7193,
+      "step": 521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000996290189612531,
+      "loss": 6.1718,
+      "step": 522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009960840890354493,
+      "loss": 6.2293,
+      "step": 523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009958779884583677,
+      "loss": 6.1392,
+      "step": 524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009956718878812861,
+      "loss": 7.8158,
+      "step": 525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0009954657873042045,
+      "loss": 6.4048,
+      "step": 526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000995259686727123,
+      "loss": 6.4098,
+      "step": 527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009950535861500413,
+      "loss": 5.6295,
+      "step": 528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009948474855729598,
+      "loss": 5.9783,
+      "step": 529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000994641384995878,
+      "loss": 6.8324,
+      "step": 530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009944352844187964,
+      "loss": 5.4508,
+      "step": 531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009942291838417148,
+      "loss": 5.4414,
+      "step": 532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009940230832646332,
+      "loss": 5.6033,
+      "step": 533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009938169826875516,
+      "loss": 5.6369,
+      "step": 534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00099361088211047,
+      "loss": 6.7861,
+      "step": 535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009934047815333884,
+      "loss": 5.7911,
+      "step": 536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009931986809563068,
+      "loss": 5.5656,
+      "step": 537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009929925803792252,
+      "loss": 6.1179,
+      "step": 538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009927864798021434,
+      "loss": 5.8315,
+      "step": 539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009925803792250618,
+      "loss": 5.8819,
+      "step": 540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009923742786479802,
+      "loss": 6.0392,
+      "step": 541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009921681780708986,
+      "loss": 5.9089,
+      "step": 542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000991962077493817,
+      "loss": 7.3658,
+      "step": 543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0009917559769167354,
+      "loss": 6.1079,
+      "step": 544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009915498763396538,
+      "loss": 5.4081,
+      "step": 545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009913437757625722,
+      "loss": 9.2302,
+      "step": 546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009911376751854906,
+      "loss": 6.5225,
+      "step": 547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009909315746084088,
+      "loss": 8.659,
+      "step": 548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009907254740313272,
+      "loss": 6.7841,
+      "step": 549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009905193734542456,
+      "loss": 5.7645,
+      "step": 550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000990313272877164,
+      "loss": 7.3645,
+      "step": 551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009901071723000825,
+      "loss": 8.1962,
+      "step": 552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009899010717230009,
+      "loss": 7.488,
+      "step": 553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009896949711459193,
+      "loss": 6.9756,
+      "step": 554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009894888705688377,
+      "loss": 6.3789,
+      "step": 555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000989282769991756,
+      "loss": 6.9698,
+      "step": 556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009890766694146743,
+      "loss": 6.4159,
+      "step": 557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009888705688375927,
+      "loss": 6.1172,
+      "step": 558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000988664468260511,
+      "loss": 5.6229,
+      "step": 559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0009884583676834295,
+      "loss": 5.76,
+      "step": 560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000988252267106348,
+      "loss": 6.4004,
+      "step": 561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009880461665292663,
+      "loss": 6.3716,
+      "step": 562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009878400659521847,
+      "loss": 5.6781,
+      "step": 563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009876339653751031,
+      "loss": 6.5148,
+      "step": 564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009874278647980215,
+      "loss": 7.1262,
+      "step": 565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009872217642209397,
+      "loss": 5.5748,
+      "step": 566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009870156636438581,
+      "loss": 6.7373,
+      "step": 567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009868095630667765,
+      "loss": 6.3384,
+      "step": 568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000986603462489695,
+      "loss": 5.9258,
+      "step": 569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009863973619126133,
+      "loss": 6.115,
+      "step": 570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009861912613355318,
+      "loss": 6.5746,
+      "step": 571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009859851607584502,
+      "loss": 5.8817,
+      "step": 572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009857790601813686,
+      "loss": 6.9406,
+      "step": 573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000985572959604287,
+      "loss": 5.98,
+      "step": 574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009853668590272052,
+      "loss": 5.7261,
+      "step": 575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009851607584501236,
+      "loss": 6.1306,
+      "step": 576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000984954657873042,
+      "loss": 6.0512,
+      "step": 577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009847485572959604,
+      "loss": 5.8131,
+      "step": 578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0009845424567188788,
+      "loss": 6.4398,
+      "step": 579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009843363561417972,
+      "loss": 6.8275,
+      "step": 580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009841302555647156,
+      "loss": 5.678,
+      "step": 581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000983924154987634,
+      "loss": 6.1303,
+      "step": 582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009837180544105524,
+      "loss": 5.928,
+      "step": 583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009835119538334708,
+      "loss": 6.2767,
+      "step": 584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000983305853256389,
+      "loss": 6.4945,
+      "step": 585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009830997526793074,
+      "loss": 5.5189,
+      "step": 586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009828936521022258,
+      "loss": 6.61,
+      "step": 587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009826875515251442,
+      "loss": 5.8178,
+      "step": 588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009824814509480626,
+      "loss": 6.6588,
+      "step": 589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000982275350370981,
+      "loss": 6.1317,
+      "step": 590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009820692497938995,
+      "loss": 5.7391,
+      "step": 591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009818631492168179,
+      "loss": 5.8252,
+      "step": 592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009816570486397363,
+      "loss": 6.0155,
+      "step": 593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009814509480626545,
+      "loss": 6.0274,
+      "step": 594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009812448474855729,
+      "loss": 5.8197,
+      "step": 595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009810387469084913,
+      "loss": 5.8393,
+      "step": 596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0009808326463314099,
+      "loss": 11.543,
+      "step": 597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000980626545754328,
+      "loss": 10.7682,
+      "step": 598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009804204451772465,
+      "loss": 5.8935,
+      "step": 599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000980214344600165,
+      "loss": 6.1574,
+      "step": 600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009800082440230833,
+      "loss": 5.7418,
+      "step": 601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009798021434460017,
+      "loss": 6.3603,
+      "step": 602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009795960428689201,
+      "loss": 5.6521,
+      "step": 603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009793899422918385,
+      "loss": 5.889,
+      "step": 604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000979183841714757,
+      "loss": 6.4798,
+      "step": 605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009789777411376753,
+      "loss": 5.6732,
+      "step": 606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009787716405605935,
+      "loss": 7.4409,
+      "step": 607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000978565539983512,
+      "loss": 6.6196,
+      "step": 608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009783594394064303,
+      "loss": 10.0377,
+      "step": 609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009781533388293487,
+      "loss": 6.0067,
+      "step": 610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009779472382522672,
+      "loss": 6.6858,
+      "step": 611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009777411376751856,
+      "loss": 5.8716,
+      "step": 612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000977535037098104,
+      "loss": 5.8858,
+      "step": 613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009773289365210224,
+      "loss": 6.4251,
+      "step": 614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0009771228359439408,
+      "loss": 6.1316,
+      "step": 615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009769167353668592,
+      "loss": 5.8224,
+      "step": 616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009767106347897774,
+      "loss": 7.0245,
+      "step": 617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009765045342126958,
+      "loss": 5.944,
+      "step": 618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009762984336356142,
+      "loss": 6.2897,
+      "step": 619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009760923330585326,
+      "loss": 7.0811,
+      "step": 620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009758862324814509,
+      "loss": 6.229,
+      "step": 621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009756801319043693,
+      "loss": 6.4282,
+      "step": 622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009754740313272877,
+      "loss": 6.1066,
+      "step": 623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009752679307502061,
+      "loss": 6.8906,
+      "step": 624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009750618301731245,
+      "loss": 5.9767,
+      "step": 625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009748557295960428,
+      "loss": 6.1151,
+      "step": 626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009746496290189612,
+      "loss": 5.8014,
+      "step": 627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009744435284418796,
+      "loss": 6.9561,
+      "step": 628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000974237427864798,
+      "loss": 7.1122,
+      "step": 629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009740313272877163,
+      "loss": 5.7136,
+      "step": 630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009738252267106347,
+      "loss": 6.04,
+      "step": 631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009736191261335531,
+      "loss": 6.5103,
+      "step": 632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0009734130255564716,
+      "loss": 6.1512,
+      "step": 633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00097320692497939,
+      "loss": 5.9868,
+      "step": 634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009730008244023083,
+      "loss": 6.0231,
+      "step": 635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009727947238252267,
+      "loss": 6.0043,
+      "step": 636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009725886232481451,
+      "loss": 5.7622,
+      "step": 637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009723825226710635,
+      "loss": 5.4486,
+      "step": 638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009721764220939819,
+      "loss": 5.8163,
+      "step": 639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009719703215169002,
+      "loss": 8.7517,
+      "step": 640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009717642209398186,
+      "loss": 7.6983,
+      "step": 641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000971558120362737,
+      "loss": 6.1003,
+      "step": 642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009713520197856554,
+      "loss": 6.0597,
+      "step": 643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009711459192085737,
+      "loss": 5.8145,
+      "step": 644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009709398186314922,
+      "loss": 5.3267,
+      "step": 645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009707337180544106,
+      "loss": 6.5171,
+      "step": 646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000970527617477329,
+      "loss": 7.5778,
+      "step": 647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009703215169002474,
+      "loss": 7.7774,
+      "step": 648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009701154163231657,
+      "loss": 6.0906,
+      "step": 649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009699093157460841,
+      "loss": 6.9724,
+      "step": 650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0009697032151690026,
+      "loss": 7.2344,
+      "step": 651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000969497114591921,
+      "loss": 5.8526,
+      "step": 652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009692910140148393,
+      "loss": 6.0832,
+      "step": 653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009690849134377577,
+      "loss": 6.0617,
+      "step": 654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009688788128606761,
+      "loss": 6.3513,
+      "step": 655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009686727122835945,
+      "loss": 5.7883,
+      "step": 656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009684666117065129,
+      "loss": 5.6316,
+      "step": 657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009682605111294312,
+      "loss": 8.4773,
+      "step": 658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009680544105523496,
+      "loss": 6.0611,
+      "step": 659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000967848309975268,
+      "loss": 6.1135,
+      "step": 660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009676422093981864,
+      "loss": 6.2206,
+      "step": 661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009674361088211047,
+      "loss": 5.983,
+      "step": 662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009672300082440231,
+      "loss": 5.6638,
+      "step": 663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009670239076669415,
+      "loss": 6.1116,
+      "step": 664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009668178070898599,
+      "loss": 5.7003,
+      "step": 665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009666117065127783,
+      "loss": 6.0617,
+      "step": 666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0009664056059356966,
+      "loss": 6.1331,
+      "step": 667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000966199505358615,
+      "loss": 6.4029,
+      "step": 668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009659934047815334,
+      "loss": 5.7595,
+      "step": 669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009657873042044518,
+      "loss": 5.7665,
+      "step": 670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009655812036273703,
+      "loss": 6.1993,
+      "step": 671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009653751030502886,
+      "loss": 6.7869,
+      "step": 672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000965169002473207,
+      "loss": 5.6794,
+      "step": 673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009649629018961254,
+      "loss": 5.5852,
+      "step": 674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009647568013190438,
+      "loss": 6.2421,
+      "step": 675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009645507007419621,
+      "loss": 5.797,
+      "step": 676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009643446001648805,
+      "loss": 5.5341,
+      "step": 677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009641384995877989,
+      "loss": 5.6733,
+      "step": 678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009639323990107173,
+      "loss": 6.861,
+      "step": 679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009637262984336357,
+      "loss": 6.9362,
+      "step": 680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000963520197856554,
+      "loss": 5.9752,
+      "step": 681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009633140972794724,
+      "loss": 5.6336,
+      "step": 682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009631079967023908,
+      "loss": 6.5689,
+      "step": 683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009629018961253092,
+      "loss": 5.8262,
+      "step": 684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009626957955482275,
+      "loss": 6.0148,
+      "step": 685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0009624896949711459,
+      "loss": 6.2406,
+      "step": 686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009622835943940643,
+      "loss": 5.7069,
+      "step": 687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009620774938169827,
+      "loss": 5.8862,
+      "step": 688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009618713932399011,
+      "loss": 6.4856,
+      "step": 689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009616652926628194,
+      "loss": 5.7372,
+      "step": 690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009614591920857378,
+      "loss": 5.6619,
+      "step": 691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009612530915086563,
+      "loss": 6.8182,
+      "step": 692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009610469909315747,
+      "loss": 6.0254,
+      "step": 693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000960840890354493,
+      "loss": 5.8925,
+      "step": 694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009606347897774114,
+      "loss": 5.8514,
+      "step": 695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009604286892003298,
+      "loss": 5.7904,
+      "step": 696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009602225886232482,
+      "loss": 5.5523,
+      "step": 697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009600164880461666,
+      "loss": 5.7803,
+      "step": 698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009598103874690849,
+      "loss": 7.3014,
+      "step": 699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009596042868920033,
+      "loss": 5.8712,
+      "step": 700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009593981863149217,
+      "loss": 5.8677,
+      "step": 701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009591920857378401,
+      "loss": 5.6388,
+      "step": 702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009589859851607585,
+      "loss": 5.9895,
+      "step": 703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0009587798845836768,
+      "loss": 7.5941,
+      "step": 704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009585737840065952,
+      "loss": 8.423,
+      "step": 705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009583676834295136,
+      "loss": 6.2247,
+      "step": 706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000958161582852432,
+      "loss": 6.7862,
+      "step": 707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009579554822753503,
+      "loss": 6.3974,
+      "step": 708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009577493816982687,
+      "loss": 5.9091,
+      "step": 709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009575432811211871,
+      "loss": 6.0214,
+      "step": 710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009573371805441055,
+      "loss": 6.0579,
+      "step": 711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000957131079967024,
+      "loss": 5.7752,
+      "step": 712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009569249793899422,
+      "loss": 6.3687,
+      "step": 713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009567188788128607,
+      "loss": 6.7535,
+      "step": 714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009565127782357791,
+      "loss": 6.2714,
+      "step": 715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009563066776586975,
+      "loss": 5.5626,
+      "step": 716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009561005770816158,
+      "loss": 5.6496,
+      "step": 717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009558944765045342,
+      "loss": 5.9608,
+      "step": 718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009556883759274526,
+      "loss": 5.8325,
+      "step": 719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000955482275350371,
+      "loss": 5.6221,
+      "step": 720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009552761747732894,
+      "loss": 6.2829,
+      "step": 721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0009550700741962077,
+      "loss": 6.12,
+      "step": 722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009548639736191261,
+      "loss": 5.766,
+      "step": 723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009546578730420445,
+      "loss": 6.9919,
+      "step": 724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009544517724649629,
+      "loss": 5.8537,
+      "step": 725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009542456718878813,
+      "loss": 6.3609,
+      "step": 726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009540395713107996,
+      "loss": 5.3722,
+      "step": 727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000953833470733718,
+      "loss": 5.931,
+      "step": 728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009536273701566364,
+      "loss": 5.5387,
+      "step": 729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009534212695795548,
+      "loss": 6.1527,
+      "step": 730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009532151690024731,
+      "loss": 5.6695,
+      "step": 731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009530090684253915,
+      "loss": 5.7496,
+      "step": 732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00095280296784831,
+      "loss": 7.9015,
+      "step": 733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009525968672712284,
+      "loss": 5.6742,
+      "step": 734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009523907666941468,
+      "loss": 6.3592,
+      "step": 735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009521846661170651,
+      "loss": 5.7121,
+      "step": 736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009519785655399835,
+      "loss": 5.6345,
+      "step": 737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000951772464962902,
+      "loss": 5.8027,
+      "step": 738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009515663643858204,
+      "loss": 6.5498,
+      "step": 739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0009513602638087387,
+      "loss": 6.0027,
+      "step": 740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009511541632316571,
+      "loss": 6.9162,
+      "step": 741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009509480626545755,
+      "loss": 5.8673,
+      "step": 742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009507419620774939,
+      "loss": 8.5893,
+      "step": 743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009505358615004123,
+      "loss": 6.5621,
+      "step": 744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009503297609233306,
+      "loss": 8.5254,
+      "step": 745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000950123660346249,
+      "loss": 5.6875,
+      "step": 746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009499175597691674,
+      "loss": 5.8806,
+      "step": 747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009497114591920858,
+      "loss": 7.409,
+      "step": 748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009495053586150041,
+      "loss": 5.4973,
+      "step": 749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009492992580379225,
+      "loss": 5.793,
+      "step": 750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009490931574608409,
+      "loss": 6.51,
+      "step": 751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009488870568837594,
+      "loss": 6.1294,
+      "step": 752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009486809563066778,
+      "loss": 5.7197,
+      "step": 753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009484748557295961,
+      "loss": 6.7428,
+      "step": 754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009482687551525145,
+      "loss": 5.8888,
+      "step": 755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009480626545754329,
+      "loss": 6.7836,
+      "step": 756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009478565539983513,
+      "loss": 7.1534,
+      "step": 757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0009476504534212697,
+      "loss": 5.9396,
+      "step": 758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000947444352844188,
+      "loss": 5.8608,
+      "step": 759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009472382522671064,
+      "loss": 5.8001,
+      "step": 760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009470321516900248,
+      "loss": 5.919,
+      "step": 761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009468260511129432,
+      "loss": 6.6078,
+      "step": 762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009466199505358615,
+      "loss": 5.6331,
+      "step": 763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009464138499587799,
+      "loss": 5.5283,
+      "step": 764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009462077493816983,
+      "loss": 5.6463,
+      "step": 765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009460016488046167,
+      "loss": 5.9516,
+      "step": 766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009457955482275351,
+      "loss": 6.3236,
+      "step": 767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009455894476504534,
+      "loss": 5.4907,
+      "step": 768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009453833470733718,
+      "loss": 6.6322,
+      "step": 769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009451772464962902,
+      "loss": 6.0348,
+      "step": 770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009449711459192086,
+      "loss": 5.9248,
+      "step": 771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009447650453421269,
+      "loss": 5.7084,
+      "step": 772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009445589447650453,
+      "loss": 5.8615,
+      "step": 773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009443528441879638,
+      "loss": 6.5772,
+      "step": 774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009441467436108822,
+      "loss": 5.7507,
+      "step": 775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0009439406430338006,
+      "loss": 5.6342,
+      "step": 776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009437345424567189,
+      "loss": 6.2213,
+      "step": 777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009435284418796373,
+      "loss": 5.8093,
+      "step": 778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009433223413025557,
+      "loss": 5.9293,
+      "step": 779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009431162407254741,
+      "loss": 5.7748,
+      "step": 780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009429101401483924,
+      "loss": 5.6932,
+      "step": 781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009427040395713108,
+      "loss": 5.9761,
+      "step": 782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009424979389942292,
+      "loss": 5.7706,
+      "step": 783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009422918384171476,
+      "loss": 5.7678,
+      "step": 784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000942085737840066,
+      "loss": 6.2639,
+      "step": 785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009418796372629843,
+      "loss": 6.5422,
+      "step": 786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009416735366859027,
+      "loss": 6.2213,
+      "step": 787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009414674361088211,
+      "loss": 6.2161,
+      "step": 788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009412613355317395,
+      "loss": 6.276,
+      "step": 789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009410552349546579,
+      "loss": 6.6628,
+      "step": 790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009408491343775762,
+      "loss": 5.846,
+      "step": 791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0009406430338004946,
+      "loss": 5.4807,
+      "step": 792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000940436933223413,
+      "loss": 5.9381,
+      "step": 793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009402308326463315,
+      "loss": 6.06,
+      "step": 794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009400247320692498,
+      "loss": 5.9511,
+      "step": 795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009398186314921682,
+      "loss": 7.8278,
+      "step": 796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009396125309150866,
+      "loss": 7.4368,
+      "step": 797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000939406430338005,
+      "loss": 5.7798,
+      "step": 798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009392003297609234,
+      "loss": 5.8919,
+      "step": 799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009389942291838417,
+      "loss": 5.8584,
+      "step": 800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009387881286067601,
+      "loss": 6.1988,
+      "step": 801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009385820280296785,
+      "loss": 6.0095,
+      "step": 802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009383759274525969,
+      "loss": 6.4251,
+      "step": 803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009381698268755152,
+      "loss": 5.9863,
+      "step": 804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009379637262984336,
+      "loss": 6.6983,
+      "step": 805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000937757625721352,
+      "loss": 6.2179,
+      "step": 806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009375515251442704,
+      "loss": 6.2726,
+      "step": 807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009373454245671888,
+      "loss": 5.8507,
+      "step": 808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009371393239901071,
+      "loss": 6.0319,
+      "step": 809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009369332234130255,
+      "loss": 6.6026,
+      "step": 810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0009367271228359439,
+      "loss": 5.8749,
+      "step": 811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009365210222588623,
+      "loss": 6.0392,
+      "step": 812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009363149216817806,
+      "loss": 6.3474,
+      "step": 813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000936108821104699,
+      "loss": 5.7689,
+      "step": 814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009359027205276175,
+      "loss": 5.8376,
+      "step": 815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009356966199505359,
+      "loss": 6.0129,
+      "step": 816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009354905193734543,
+      "loss": 5.9773,
+      "step": 817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009352844187963726,
+      "loss": 5.961,
+      "step": 818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000935078318219291,
+      "loss": 6.2443,
+      "step": 819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009348722176422094,
+      "loss": 5.8147,
+      "step": 820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009346661170651278,
+      "loss": 5.9247,
+      "step": 821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009344600164880462,
+      "loss": 5.8179,
+      "step": 822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009342539159109645,
+      "loss": 7.1148,
+      "step": 823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009340478153338829,
+      "loss": 5.9162,
+      "step": 824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009338417147568013,
+      "loss": 6.0271,
+      "step": 825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009336356141797197,
+      "loss": 6.1079,
+      "step": 826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000933429513602638,
+      "loss": 6.6895,
+      "step": 827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009332234130255564,
+      "loss": 5.7959,
+      "step": 828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0009330173124484748,
+      "loss": 6.6905,
+      "step": 829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009328112118713932,
+      "loss": 6.1454,
+      "step": 830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009326051112943117,
+      "loss": 6.0635,
+      "step": 831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00093239901071723,
+      "loss": 6.4842,
+      "step": 832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009321929101401484,
+      "loss": 6.0337,
+      "step": 833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009319868095630669,
+      "loss": 5.253,
+      "step": 834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009317807089859853,
+      "loss": 5.8837,
+      "step": 835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009315746084089036,
+      "loss": 5.8087,
+      "step": 836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000931368507831822,
+      "loss": 5.8084,
+      "step": 837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009311624072547404,
+      "loss": 5.5533,
+      "step": 838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009309563066776588,
+      "loss": 5.7599,
+      "step": 839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009307502061005772,
+      "loss": 6.1205,
+      "step": 840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009305441055234955,
+      "loss": 7.3542,
+      "step": 841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009303380049464139,
+      "loss": 5.7548,
+      "step": 842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009301319043693323,
+      "loss": 5.5901,
+      "step": 843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009299258037922507,
+      "loss": 6.2253,
+      "step": 844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009297197032151691,
+      "loss": 6.0265,
+      "step": 845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009295136026380874,
+      "loss": 5.5213,
+      "step": 846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0009293075020610058,
+      "loss": 5.5142,
+      "step": 847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009291014014839242,
+      "loss": 6.063,
+      "step": 848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009288953009068426,
+      "loss": 5.8205,
+      "step": 849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009286892003297609,
+      "loss": 7.4822,
+      "step": 850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009284830997526793,
+      "loss": 5.928,
+      "step": 851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009282769991755977,
+      "loss": 5.852,
+      "step": 852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009280708985985161,
+      "loss": 6.3175,
+      "step": 853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009278647980214346,
+      "loss": 6.0396,
+      "step": 854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009276586974443529,
+      "loss": 5.8648,
+      "step": 855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009274525968672713,
+      "loss": 6.1074,
+      "step": 856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009272464962901897,
+      "loss": 5.7053,
+      "step": 857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009270403957131081,
+      "loss": 5.8018,
+      "step": 858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009268342951360264,
+      "loss": 6.1271,
+      "step": 859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009266281945589448,
+      "loss": 6.2124,
+      "step": 860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009264220939818632,
+      "loss": 5.507,
+      "step": 861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009262159934047816,
+      "loss": 5.8833,
+      "step": 862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009260098928277,
+      "loss": 6.3893,
+      "step": 863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009258037922506183,
+      "loss": 5.5791,
+      "step": 864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0009255976916735367,
+      "loss": 6.1188,
+      "step": 865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009253915910964551,
+      "loss": 7.5054,
+      "step": 866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009251854905193735,
+      "loss": 8.2277,
+      "step": 867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009249793899422918,
+      "loss": 6.2631,
+      "step": 868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009247732893652102,
+      "loss": 5.8057,
+      "step": 869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009245671887881286,
+      "loss": 5.7185,
+      "step": 870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000924361088211047,
+      "loss": 5.6575,
+      "step": 871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009241549876339654,
+      "loss": 6.0452,
+      "step": 872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009239488870568837,
+      "loss": 6.3632,
+      "step": 873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009237427864798021,
+      "loss": 6.2171,
+      "step": 874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009235366859027206,
+      "loss": 6.0166,
+      "step": 875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000923330585325639,
+      "loss": 6.7128,
+      "step": 876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009231244847485574,
+      "loss": 6.1739,
+      "step": 877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009229183841714757,
+      "loss": 5.699,
+      "step": 878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009227122835943941,
+      "loss": 5.7356,
+      "step": 879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009225061830173125,
+      "loss": 5.7244,
+      "step": 880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009223000824402309,
+      "loss": 5.8828,
+      "step": 881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009220939818631492,
+      "loss": 5.7031,
+      "step": 882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0009218878812860676,
+      "loss": 6.1956,
+      "step": 883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000921681780708986,
+      "loss": 7.1721,
+      "step": 884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009214756801319044,
+      "loss": 6.5536,
+      "step": 885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009212695795548228,
+      "loss": 6.2362,
+      "step": 886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009210634789777411,
+      "loss": 6.18,
+      "step": 887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009208573784006595,
+      "loss": 5.654,
+      "step": 888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009206512778235779,
+      "loss": 7.0099,
+      "step": 889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009204451772464963,
+      "loss": 6.0267,
+      "step": 890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009202390766694146,
+      "loss": 5.7942,
+      "step": 891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000920032976092333,
+      "loss": 5.8989,
+      "step": 892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009198268755152514,
+      "loss": 6.5108,
+      "step": 893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009196207749381698,
+      "loss": 5.6047,
+      "step": 894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009194146743610883,
+      "loss": 5.675,
+      "step": 895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009192085737840066,
+      "loss": 6.5949,
+      "step": 896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000919002473206925,
+      "loss": 5.5905,
+      "step": 897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009187963726298434,
+      "loss": 7.1908,
+      "step": 898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009185902720527618,
+      "loss": 7.1361,
+      "step": 899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0009183841714756801,
+      "loss": 6.4069,
+      "step": 900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009181780708985985,
+      "loss": 6.2917,
+      "step": 901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009179719703215169,
+      "loss": 6.6934,
+      "step": 902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009177658697444353,
+      "loss": 6.4055,
+      "step": 903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009175597691673537,
+      "loss": 5.6773,
+      "step": 904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000917353668590272,
+      "loss": 6.2447,
+      "step": 905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009171475680131904,
+      "loss": 6.3167,
+      "step": 906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009169414674361088,
+      "loss": 6.228,
+      "step": 907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009167353668590272,
+      "loss": 5.6141,
+      "step": 908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009165292662819456,
+      "loss": 6.1728,
+      "step": 909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009163231657048639,
+      "loss": 5.7129,
+      "step": 910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009161170651277823,
+      "loss": 6.1262,
+      "step": 911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009159109645507007,
+      "loss": 5.6343,
+      "step": 912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009157048639736191,
+      "loss": 6.4345,
+      "step": 913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009154987633965374,
+      "loss": 5.753,
+      "step": 914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009152926628194558,
+      "loss": 6.3345,
+      "step": 915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009150865622423743,
+      "loss": 5.857,
+      "step": 916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009148804616652927,
+      "loss": 5.7666,
+      "step": 917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0009146743610882111,
+      "loss": 7.5008,
+      "step": 918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009144682605111294,
+      "loss": 6.4059,
+      "step": 919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009142621599340478,
+      "loss": 5.6591,
+      "step": 920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009140560593569662,
+      "loss": 5.7848,
+      "step": 921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009138499587798846,
+      "loss": 5.5911,
+      "step": 922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009136438582028029,
+      "loss": 5.5688,
+      "step": 923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009134377576257213,
+      "loss": 7.1468,
+      "step": 924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009132316570486398,
+      "loss": 5.9479,
+      "step": 925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009130255564715582,
+      "loss": 6.7228,
+      "step": 926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009128194558944766,
+      "loss": 5.7739,
+      "step": 927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009126133553173949,
+      "loss": 5.6202,
+      "step": 928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009124072547403133,
+      "loss": 5.5174,
+      "step": 929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009122011541632317,
+      "loss": 5.7537,
+      "step": 930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009119950535861501,
+      "loss": 6.281,
+      "step": 931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009117889530090684,
+      "loss": 6.4887,
+      "step": 932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009115828524319868,
+      "loss": 5.7442,
+      "step": 933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009113767518549052,
+      "loss": 5.8476,
+      "step": 934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009111706512778237,
+      "loss": 6.0147,
+      "step": 935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0009109645507007421,
+      "loss": 5.898,
+      "step": 936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009107584501236604,
+      "loss": 5.9775,
+      "step": 937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009105523495465788,
+      "loss": 5.4357,
+      "step": 938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009103462489694972,
+      "loss": 5.5592,
+      "step": 939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009101401483924156,
+      "loss": 5.6846,
+      "step": 940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000909934047815334,
+      "loss": 6.6206,
+      "step": 941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009097279472382523,
+      "loss": 6.1468,
+      "step": 942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009095218466611707,
+      "loss": 7.4961,
+      "step": 943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009093157460840891,
+      "loss": 6.2355,
+      "step": 944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009091096455070075,
+      "loss": 6.0147,
+      "step": 945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009089035449299258,
+      "loss": 5.7341,
+      "step": 946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009086974443528442,
+      "loss": 5.7853,
+      "step": 947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009084913437757626,
+      "loss": 5.7964,
+      "step": 948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000908285243198681,
+      "loss": 11.4728,
+      "step": 949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009080791426215994,
+      "loss": 5.3236,
+      "step": 950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009078730420445177,
+      "loss": 6.0453,
+      "step": 951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009076669414674361,
+      "loss": 5.9395,
+      "step": 952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0009074608408903545,
+      "loss": 5.9022,
+      "step": 953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000907254740313273,
+      "loss": 5.344,
+      "step": 954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009070486397361912,
+      "loss": 6.1423,
+      "step": 955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009068425391591097,
+      "loss": 5.8457,
+      "step": 956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009066364385820281,
+      "loss": 6.2337,
+      "step": 957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009064303380049465,
+      "loss": 5.6158,
+      "step": 958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009062242374278649,
+      "loss": 6.8799,
+      "step": 959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009060181368507832,
+      "loss": 5.7202,
+      "step": 960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009058120362737016,
+      "loss": 5.5842,
+      "step": 961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00090560593569662,
+      "loss": 6.0909,
+      "step": 962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009053998351195384,
+      "loss": 6.9954,
+      "step": 963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009051937345424568,
+      "loss": 5.7931,
+      "step": 964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009049876339653751,
+      "loss": 5.5714,
+      "step": 965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009047815333882935,
+      "loss": 5.5633,
+      "step": 966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009045754328112119,
+      "loss": 6.6389,
+      "step": 967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009043693322341303,
+      "loss": 5.7287,
+      "step": 968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009041632316570486,
+      "loss": 5.7334,
+      "step": 969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000903957131079967,
+      "loss": 5.4841,
+      "step": 970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009037510305028854,
+      "loss": 5.7807,
+      "step": 971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0009035449299258038,
+      "loss": 5.5854,
+      "step": 972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009033388293487222,
+      "loss": 5.8714,
+      "step": 973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009031327287716405,
+      "loss": 8.4749,
+      "step": 974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000902926628194559,
+      "loss": 5.8645,
+      "step": 975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009027205276174774,
+      "loss": 6.2303,
+      "step": 976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009025144270403958,
+      "loss": 6.5445,
+      "step": 977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009023083264633141,
+      "loss": 5.7156,
+      "step": 978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009021022258862325,
+      "loss": 6.167,
+      "step": 979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009018961253091509,
+      "loss": 5.6918,
+      "step": 980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009016900247320693,
+      "loss": 5.2162,
+      "step": 981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009014839241549877,
+      "loss": 5.4247,
+      "step": 982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000901277823577906,
+      "loss": 6.0862,
+      "step": 983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009010717230008244,
+      "loss": 5.7384,
+      "step": 984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009008656224237428,
+      "loss": 5.2791,
+      "step": 985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009006595218466612,
+      "loss": 5.8997,
+      "step": 986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009004534212695795,
+      "loss": 6.15,
+      "step": 987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009002473206924979,
+      "loss": 5.6691,
+      "step": 988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0009000412201154163,
+      "loss": 5.9815,
+      "step": 989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0008998351195383347,
+      "loss": 5.3707,
+      "step": 990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008996290189612531,
+      "loss": 5.94,
+      "step": 991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008994229183841714,
+      "loss": 6.3036,
+      "step": 992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008992168178070898,
+      "loss": 5.7854,
+      "step": 993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008990107172300082,
+      "loss": 6.2649,
+      "step": 994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008988046166529266,
+      "loss": 5.6664,
+      "step": 995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000898598516075845,
+      "loss": 5.0298,
+      "step": 996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008983924154987634,
+      "loss": 5.5162,
+      "step": 997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008981863149216818,
+      "loss": 6.4971,
+      "step": 998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008979802143446002,
+      "loss": 6.0985,
+      "step": 999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008977741137675186,
+      "loss": 5.6351,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 6.1492109298706055,
+      "eval_runtime": 1252.2007,
+      "eval_samples_per_second": 2.11,
+      "eval_steps_per_second": 0.264,
+      "eval_wer": 2.0010115033716778,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008975680131904369,
+      "loss": 6.5024,
+      "step": 1001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008973619126133553,
+      "loss": 5.7955,
+      "step": 1002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008971558120362737,
+      "loss": 5.8845,
+      "step": 1003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008969497114591921,
+      "loss": 6.2277,
+      "step": 1004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008967436108821105,
+      "loss": 5.9537,
+      "step": 1005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008965375103050288,
+      "loss": 5.8075,
+      "step": 1006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0008963314097279472,
+      "loss": 6.9245,
+      "step": 1007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008961253091508656,
+      "loss": 6.3975,
+      "step": 1008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000895919208573784,
+      "loss": 6.2743,
+      "step": 1009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008957131079967023,
+      "loss": 5.6112,
+      "step": 1010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008955070074196207,
+      "loss": 5.7932,
+      "step": 1011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008953009068425391,
+      "loss": 5.9059,
+      "step": 1012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008950948062654575,
+      "loss": 5.6093,
+      "step": 1013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008948887056883759,
+      "loss": 5.7219,
+      "step": 1014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008946826051112942,
+      "loss": 6.5035,
+      "step": 1015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008944765045342126,
+      "loss": 5.5499,
+      "step": 1016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000894270403957131,
+      "loss": 5.8064,
+      "step": 1017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008940643033800496,
+      "loss": 5.6278,
+      "step": 1018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008938582028029679,
+      "loss": 6.9634,
+      "step": 1019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008936521022258863,
+      "loss": 6.1874,
+      "step": 1020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008934460016488047,
+      "loss": 5.6973,
+      "step": 1021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008932399010717231,
+      "loss": 6.163,
+      "step": 1022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008930338004946415,
+      "loss": 5.7016,
+      "step": 1023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008928276999175598,
+      "loss": 6.0438,
+      "step": 1024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0008926215993404782,
+      "loss": 6.1569,
+      "step": 1025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008924154987633966,
+      "loss": 6.1029,
+      "step": 1026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000892209398186315,
+      "loss": 5.8521,
+      "step": 1027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008920032976092334,
+      "loss": 6.1279,
+      "step": 1028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008917971970321517,
+      "loss": 5.9389,
+      "step": 1029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008915910964550701,
+      "loss": 5.6962,
+      "step": 1030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008913849958779885,
+      "loss": 6.1379,
+      "step": 1031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008911788953009069,
+      "loss": 5.9054,
+      "step": 1032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008909727947238252,
+      "loss": 5.5586,
+      "step": 1033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008907666941467436,
+      "loss": 5.5209,
+      "step": 1034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000890560593569662,
+      "loss": 5.8775,
+      "step": 1035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008903544929925805,
+      "loss": 6.328,
+      "step": 1036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008901483924154989,
+      "loss": 5.5212,
+      "step": 1037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008899422918384172,
+      "loss": 5.8285,
+      "step": 1038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008897361912613356,
+      "loss": 6.4879,
+      "step": 1039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000889530090684254,
+      "loss": 6.6968,
+      "step": 1040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008893239901071724,
+      "loss": 6.5249,
+      "step": 1041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008891178895300907,
+      "loss": 5.6785,
+      "step": 1042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0008889117889530091,
+      "loss": 5.7217,
+      "step": 1043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008887056883759275,
+      "loss": 6.7814,
+      "step": 1044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008884995877988459,
+      "loss": 6.9561,
+      "step": 1045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008882934872217643,
+      "loss": 5.9531,
+      "step": 1046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008880873866446826,
+      "loss": 6.0275,
+      "step": 1047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000887881286067601,
+      "loss": 5.6911,
+      "step": 1048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008876751854905194,
+      "loss": 5.8849,
+      "step": 1049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008874690849134378,
+      "loss": 6.4503,
+      "step": 1050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008872629843363561,
+      "loss": 6.828,
+      "step": 1051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008870568837592745,
+      "loss": 5.6407,
+      "step": 1052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008868507831821929,
+      "loss": 6.2972,
+      "step": 1053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008866446826051113,
+      "loss": 5.7767,
+      "step": 1054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008864385820280297,
+      "loss": 6.1377,
+      "step": 1055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000886232481450948,
+      "loss": 5.6328,
+      "step": 1056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008860263808738665,
+      "loss": 6.4933,
+      "step": 1057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008858202802967849,
+      "loss": 5.6022,
+      "step": 1058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008856141797197033,
+      "loss": 6.4027,
+      "step": 1059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0008854080791426217,
+      "loss": 5.9105,
+      "step": 1060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00088520197856554,
+      "loss": 6.1077,
+      "step": 1061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008849958779884584,
+      "loss": 5.7963,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008847897774113768,
+      "loss": 6.2369,
+      "step": 1063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008845836768342952,
+      "loss": 5.5735,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008843775762572135,
+      "loss": 5.9389,
+      "step": 1065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008841714756801319,
+      "loss": 5.6194,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008839653751030503,
+      "loss": 5.8619,
+      "step": 1067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008837592745259687,
+      "loss": 5.6772,
+      "step": 1068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008835531739488871,
+      "loss": 5.9223,
+      "step": 1069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008833470733718054,
+      "loss": 6.0648,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008831409727947238,
+      "loss": 5.6514,
+      "step": 1071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008829348722176422,
+      "loss": 5.396,
+      "step": 1072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008827287716405606,
+      "loss": 5.681,
+      "step": 1073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008825226710634789,
+      "loss": 6.0192,
+      "step": 1074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008823165704863973,
+      "loss": 6.0451,
+      "step": 1075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008821104699093157,
+      "loss": 6.5253,
+      "step": 1076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008819043693322342,
+      "loss": 5.9291,
+      "step": 1077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008816982687551526,
+      "loss": 5.5373,
+      "step": 1078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0008814921681780709,
+      "loss": 5.9191,
+      "step": 1079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008812860676009893,
+      "loss": 6.284,
+      "step": 1080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008810799670239077,
+      "loss": 6.4584,
+      "step": 1081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008808738664468261,
+      "loss": 6.4029,
+      "step": 1082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008806677658697445,
+      "loss": 6.0231,
+      "step": 1083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008804616652926628,
+      "loss": 5.742,
+      "step": 1084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008802555647155812,
+      "loss": 5.4892,
+      "step": 1085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008800494641384996,
+      "loss": 6.3841,
+      "step": 1086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000879843363561418,
+      "loss": 6.9114,
+      "step": 1087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008796372629843363,
+      "loss": 6.3513,
+      "step": 1088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008794311624072547,
+      "loss": 6.561,
+      "step": 1089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008792250618301731,
+      "loss": 5.515,
+      "step": 1090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008790189612530915,
+      "loss": 5.9804,
+      "step": 1091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008788128606760099,
+      "loss": 6.0353,
+      "step": 1092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008786067600989282,
+      "loss": 5.6856,
+      "step": 1093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008784006595218466,
+      "loss": 5.8339,
+      "step": 1094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000878194558944765,
+      "loss": 5.6471,
+      "step": 1095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008779884583676834,
+      "loss": 6.5953,
+      "step": 1096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0008777823577906017,
+      "loss": 6.6594,
+      "step": 1097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008775762572135202,
+      "loss": 7.3791,
+      "step": 1098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008773701566364386,
+      "loss": 5.7074,
+      "step": 1099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000877164056059357,
+      "loss": 5.5945,
+      "step": 1100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008769579554822754,
+      "loss": 6.1098,
+      "step": 1101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008767518549051937,
+      "loss": 5.8439,
+      "step": 1102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008765457543281121,
+      "loss": 5.7616,
+      "step": 1103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008763396537510305,
+      "loss": 5.4494,
+      "step": 1104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008761335531739489,
+      "loss": 5.455,
+      "step": 1105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008759274525968672,
+      "loss": 5.3426,
+      "step": 1106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008757213520197856,
+      "loss": 5.6139,
+      "step": 1107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000875515251442704,
+      "loss": 5.5774,
+      "step": 1108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008753091508656224,
+      "loss": 5.6453,
+      "step": 1109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008751030502885409,
+      "loss": 6.1365,
+      "step": 1110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008748969497114591,
+      "loss": 5.7588,
+      "step": 1111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008746908491343776,
+      "loss": 5.8667,
+      "step": 1112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000874484748557296,
+      "loss": 5.8677,
+      "step": 1113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008742786479802144,
+      "loss": 6.256,
+      "step": 1114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0008740725474031328,
+      "loss": 6.6167,
+      "step": 1115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008738664468260511,
+      "loss": 5.4481,
+      "step": 1116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008736603462489696,
+      "loss": 6.0058,
+      "step": 1117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000873454245671888,
+      "loss": 5.7767,
+      "step": 1118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008732481450948064,
+      "loss": 5.6444,
+      "step": 1119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008730420445177247,
+      "loss": 6.307,
+      "step": 1120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008728359439406431,
+      "loss": 5.6836,
+      "step": 1121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008726298433635615,
+      "loss": 5.5824,
+      "step": 1122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008724237427864799,
+      "loss": 5.5895,
+      "step": 1123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008722176422093983,
+      "loss": 5.3565,
+      "step": 1124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008720115416323166,
+      "loss": 6.0187,
+      "step": 1125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000871805441055235,
+      "loss": 5.602,
+      "step": 1126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008715993404781534,
+      "loss": 5.3686,
+      "step": 1127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008713932399010718,
+      "loss": 6.1128,
+      "step": 1128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008711871393239901,
+      "loss": 5.5046,
+      "step": 1129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008709810387469085,
+      "loss": 6.3842,
+      "step": 1130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008707749381698269,
+      "loss": 5.8281,
+      "step": 1131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0008705688375927453,
+      "loss": 5.6772,
+      "step": 1132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008703627370156637,
+      "loss": 5.9113,
+      "step": 1133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000870156636438582,
+      "loss": 5.5275,
+      "step": 1134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008699505358615004,
+      "loss": 5.8943,
+      "step": 1135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008697444352844188,
+      "loss": 5.3634,
+      "step": 1136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008695383347073373,
+      "loss": 6.1428,
+      "step": 1137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008693322341302556,
+      "loss": 5.7043,
+      "step": 1138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000869126133553174,
+      "loss": 5.3332,
+      "step": 1139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008689200329760924,
+      "loss": 5.4223,
+      "step": 1140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008687139323990108,
+      "loss": 5.6035,
+      "step": 1141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008685078318219292,
+      "loss": 6.0933,
+      "step": 1142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008683017312448475,
+      "loss": 6.476,
+      "step": 1143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008680956306677659,
+      "loss": 5.7891,
+      "step": 1144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008678895300906843,
+      "loss": 5.8356,
+      "step": 1145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008676834295136027,
+      "loss": 6.3903,
+      "step": 1146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008674773289365211,
+      "loss": 5.7805,
+      "step": 1147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008672712283594394,
+      "loss": 5.7685,
+      "step": 1148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008670651277823578,
+      "loss": 7.177,
+      "step": 1149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0008668590272052762,
+      "loss": 5.9728,
+      "step": 1150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008666529266281946,
+      "loss": 6.5439,
+      "step": 1151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008664468260511129,
+      "loss": 6.3413,
+      "step": 1152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008662407254740313,
+      "loss": 5.6954,
+      "step": 1153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008660346248969497,
+      "loss": 5.8025,
+      "step": 1154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008658285243198681,
+      "loss": 5.9334,
+      "step": 1155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008656224237427865,
+      "loss": 5.7321,
+      "step": 1156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008654163231657048,
+      "loss": 5.8495,
+      "step": 1157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008652102225886233,
+      "loss": 6.3292,
+      "step": 1158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008650041220115417,
+      "loss": 5.38,
+      "step": 1159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008647980214344601,
+      "loss": 5.8129,
+      "step": 1160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008645919208573784,
+      "loss": 5.6957,
+      "step": 1161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008643858202802968,
+      "loss": 5.5909,
+      "step": 1162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008641797197032152,
+      "loss": 5.7853,
+      "step": 1163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008639736191261336,
+      "loss": 5.3367,
+      "step": 1164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000863767518549052,
+      "loss": 5.536,
+      "step": 1165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008635614179719703,
+      "loss": 5.9351,
+      "step": 1166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008633553173948887,
+      "loss": 5.9232,
+      "step": 1167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0008631492168178071,
+      "loss": 5.2118,
+      "step": 1168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008629431162407255,
+      "loss": 5.6374,
+      "step": 1169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008627370156636439,
+      "loss": 5.9194,
+      "step": 1170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008625309150865622,
+      "loss": 6.015,
+      "step": 1171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008623248145094806,
+      "loss": 5.3158,
+      "step": 1172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000862118713932399,
+      "loss": 5.9325,
+      "step": 1173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008619126133553174,
+      "loss": 5.3442,
+      "step": 1174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008617065127782357,
+      "loss": 5.8339,
+      "step": 1175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008615004122011541,
+      "loss": 6.7221,
+      "step": 1176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008612943116240725,
+      "loss": 5.8754,
+      "step": 1177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000861088211046991,
+      "loss": 5.6991,
+      "step": 1178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008608821104699094,
+      "loss": 5.6825,
+      "step": 1179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008606760098928277,
+      "loss": 5.7399,
+      "step": 1180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008604699093157461,
+      "loss": 5.5135,
+      "step": 1181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008602638087386645,
+      "loss": 6.2864,
+      "step": 1182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008600577081615829,
+      "loss": 5.8132,
+      "step": 1183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008598516075845012,
+      "loss": 5.4415,
+      "step": 1184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0008596455070074196,
+      "loss": 6.3479,
+      "step": 1185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000859439406430338,
+      "loss": 5.9894,
+      "step": 1186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008592333058532564,
+      "loss": 5.9649,
+      "step": 1187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008590272052761748,
+      "loss": 5.8121,
+      "step": 1188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008588211046990931,
+      "loss": 5.7591,
+      "step": 1189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008586150041220115,
+      "loss": 6.2276,
+      "step": 1190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008584089035449299,
+      "loss": 5.5854,
+      "step": 1191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008582028029678483,
+      "loss": 5.9735,
+      "step": 1192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008579967023907666,
+      "loss": 6.1495,
+      "step": 1193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000857790601813685,
+      "loss": 5.7988,
+      "step": 1194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008575845012366034,
+      "loss": 5.8238,
+      "step": 1195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008573784006595218,
+      "loss": 5.5216,
+      "step": 1196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008571723000824402,
+      "loss": 5.8972,
+      "step": 1197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008569661995053585,
+      "loss": 5.675,
+      "step": 1198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000856760098928277,
+      "loss": 5.6258,
+      "step": 1199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008565539983511954,
+      "loss": 5.2509,
+      "step": 1200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008563478977741138,
+      "loss": 6.0708,
+      "step": 1201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008561417971970322,
+      "loss": 5.813,
+      "step": 1202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0008559356966199505,
+      "loss": 5.7208,
+      "step": 1203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000855729596042869,
+      "loss": 5.4246,
+      "step": 1204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008555234954657874,
+      "loss": 5.4555,
+      "step": 1205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008553173948887058,
+      "loss": 5.5899,
+      "step": 1206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008551112943116241,
+      "loss": 6.3723,
+      "step": 1207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008549051937345425,
+      "loss": 6.0042,
+      "step": 1208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008546990931574609,
+      "loss": 6.0567,
+      "step": 1209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008544929925803793,
+      "loss": 5.6095,
+      "step": 1210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008542868920032977,
+      "loss": 5.6663,
+      "step": 1211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000854080791426216,
+      "loss": 5.8827,
+      "step": 1212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008538746908491344,
+      "loss": 5.5519,
+      "step": 1213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008536685902720528,
+      "loss": 5.9668,
+      "step": 1214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008534624896949712,
+      "loss": 5.6683,
+      "step": 1215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008532563891178895,
+      "loss": 5.6001,
+      "step": 1216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008530502885408079,
+      "loss": 5.6482,
+      "step": 1217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008528441879637264,
+      "loss": 5.2361,
+      "step": 1218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008526380873866448,
+      "loss": 5.4159,
+      "step": 1219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008524319868095632,
+      "loss": 5.7627,
+      "step": 1220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008522258862324815,
+      "loss": 5.5442,
+      "step": 1221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0008520197856553999,
+      "loss": 5.8793,
+      "step": 1222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008518136850783183,
+      "loss": 5.6433,
+      "step": 1223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008516075845012367,
+      "loss": 5.9699,
+      "step": 1224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000851401483924155,
+      "loss": 6.3155,
+      "step": 1225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008511953833470734,
+      "loss": 5.6107,
+      "step": 1226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008509892827699918,
+      "loss": 5.4555,
+      "step": 1227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008507831821929102,
+      "loss": 5.7556,
+      "step": 1228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008505770816158286,
+      "loss": 5.425,
+      "step": 1229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008503709810387469,
+      "loss": 6.2961,
+      "step": 1230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008501648804616653,
+      "loss": 6.0284,
+      "step": 1231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008499587798845837,
+      "loss": 6.0043,
+      "step": 1232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008497526793075021,
+      "loss": 5.8501,
+      "step": 1233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008495465787304205,
+      "loss": 5.4706,
+      "step": 1234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008493404781533388,
+      "loss": 6.0526,
+      "step": 1235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008491343775762572,
+      "loss": 5.4186,
+      "step": 1236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008489282769991756,
+      "loss": 5.9058,
+      "step": 1237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000848722176422094,
+      "loss": 5.4224,
+      "step": 1238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0008485160758450124,
+      "loss": 6.2148,
+      "step": 1239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008483099752679308,
+      "loss": 6.1515,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008481038746908492,
+      "loss": 5.5251,
+      "step": 1241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008478977741137676,
+      "loss": 6.0468,
+      "step": 1242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000847691673536686,
+      "loss": 5.947,
+      "step": 1243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008474855729596043,
+      "loss": 5.5033,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008472794723825227,
+      "loss": 5.1342,
+      "step": 1245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008470733718054411,
+      "loss": 5.6845,
+      "step": 1246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008468672712283595,
+      "loss": 5.6728,
+      "step": 1247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008466611706512778,
+      "loss": 5.5753,
+      "step": 1248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008464550700741962,
+      "loss": 5.5772,
+      "step": 1249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008462489694971146,
+      "loss": 5.9385,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000846042868920033,
+      "loss": 5.9519,
+      "step": 1251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008458367683429514,
+      "loss": 6.1942,
+      "step": 1252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008456306677658697,
+      "loss": 5.8612,
+      "step": 1253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008454245671887881,
+      "loss": 5.6507,
+      "step": 1254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008452184666117065,
+      "loss": 6.2835,
+      "step": 1255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008450123660346249,
+      "loss": 5.6768,
+      "step": 1256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0008448062654575432,
+      "loss": 5.4496,
+      "step": 1257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008446001648804616,
+      "loss": 5.4771,
+      "step": 1258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00084439406430338,
+      "loss": 5.5767,
+      "step": 1259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008441879637262985,
+      "loss": 6.212,
+      "step": 1260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008439818631492169,
+      "loss": 5.8002,
+      "step": 1261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008437757625721352,
+      "loss": 5.8456,
+      "step": 1262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008435696619950536,
+      "loss": 5.7318,
+      "step": 1263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000843363561417972,
+      "loss": 5.433,
+      "step": 1264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008431574608408904,
+      "loss": 6.1216,
+      "step": 1265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008429513602638088,
+      "loss": 5.8157,
+      "step": 1266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008427452596867271,
+      "loss": 6.1754,
+      "step": 1267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008425391591096455,
+      "loss": 5.4214,
+      "step": 1268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008423330585325639,
+      "loss": 5.6013,
+      "step": 1269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008421269579554823,
+      "loss": 5.8763,
+      "step": 1270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008419208573784006,
+      "loss": 5.8299,
+      "step": 1271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000841714756801319,
+      "loss": 5.5678,
+      "step": 1272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008415086562242374,
+      "loss": 5.6819,
+      "step": 1273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008413025556471558,
+      "loss": 5.6345,
+      "step": 1274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0008410964550700742,
+      "loss": 5.583,
+      "step": 1275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008408903544929925,
+      "loss": 5.9524,
+      "step": 1276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008406842539159109,
+      "loss": 5.8557,
+      "step": 1277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008404781533388293,
+      "loss": 6.3414,
+      "step": 1278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008402720527617478,
+      "loss": 6.0703,
+      "step": 1279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000840065952184666,
+      "loss": 5.6424,
+      "step": 1280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008398598516075845,
+      "loss": 5.4155,
+      "step": 1281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008396537510305029,
+      "loss": 5.4127,
+      "step": 1282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008394476504534213,
+      "loss": 5.9167,
+      "step": 1283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008392415498763397,
+      "loss": 5.3472,
+      "step": 1284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000839035449299258,
+      "loss": 5.4642,
+      "step": 1285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008388293487221764,
+      "loss": 5.7275,
+      "step": 1286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008386232481450948,
+      "loss": 5.8869,
+      "step": 1287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008384171475680132,
+      "loss": 5.3807,
+      "step": 1288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008382110469909316,
+      "loss": 5.5301,
+      "step": 1289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008380049464138499,
+      "loss": 5.2171,
+      "step": 1290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008377988458367683,
+      "loss": 5.7887,
+      "step": 1291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008375927452596867,
+      "loss": 6.7826,
+      "step": 1292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0008373866446826051,
+      "loss": 7.5032,
+      "step": 1293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008371805441055234,
+      "loss": 5.2918,
+      "step": 1294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008369744435284418,
+      "loss": 5.299,
+      "step": 1295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008367683429513602,
+      "loss": 5.3997,
+      "step": 1296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008365622423742787,
+      "loss": 5.1969,
+      "step": 1297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008363561417971972,
+      "loss": 4.997,
+      "step": 1298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008361500412201155,
+      "loss": 7.1958,
+      "step": 1299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008359439406430339,
+      "loss": 5.1217,
+      "step": 1300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008357378400659523,
+      "loss": 6.4683,
+      "step": 1301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008355317394888707,
+      "loss": 6.3148,
+      "step": 1302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000835325638911789,
+      "loss": 5.6597,
+      "step": 1303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008351195383347074,
+      "loss": 5.8385,
+      "step": 1304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008349134377576258,
+      "loss": 5.8835,
+      "step": 1305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008347073371805442,
+      "loss": 5.7553,
+      "step": 1306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008345012366034626,
+      "loss": 5.9277,
+      "step": 1307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008342951360263809,
+      "loss": 5.4568,
+      "step": 1308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008340890354492993,
+      "loss": 6.0285,
+      "step": 1309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008338829348722177,
+      "loss": 5.4946,
+      "step": 1310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0008336768342951361,
+      "loss": 5.3608,
+      "step": 1311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008334707337180544,
+      "loss": 5.7254,
+      "step": 1312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008332646331409728,
+      "loss": 5.4738,
+      "step": 1313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008330585325638912,
+      "loss": 5.3438,
+      "step": 1314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008328524319868096,
+      "loss": 6.1135,
+      "step": 1315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000832646331409728,
+      "loss": 5.9388,
+      "step": 1316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008324402308326463,
+      "loss": 5.3091,
+      "step": 1317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008322341302555647,
+      "loss": 6.2014,
+      "step": 1318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008320280296784832,
+      "loss": 5.3778,
+      "step": 1319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008318219291014016,
+      "loss": 5.3689,
+      "step": 1320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00083161582852432,
+      "loss": 5.106,
+      "step": 1321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008314097279472383,
+      "loss": 5.3864,
+      "step": 1322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008312036273701567,
+      "loss": 5.7241,
+      "step": 1323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008309975267930751,
+      "loss": 5.5308,
+      "step": 1324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008307914262159935,
+      "loss": 5.6391,
+      "step": 1325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008305853256389118,
+      "loss": 5.5302,
+      "step": 1326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008303792250618302,
+      "loss": 5.1797,
+      "step": 1327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0008301731244847486,
+      "loss": 6.0267,
+      "step": 1328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000829967023907667,
+      "loss": 6.0694,
+      "step": 1329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008297609233305854,
+      "loss": 6.175,
+      "step": 1330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008295548227535037,
+      "loss": 5.4023,
+      "step": 1331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008293487221764221,
+      "loss": 5.5027,
+      "step": 1332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008291426215993405,
+      "loss": 5.9912,
+      "step": 1333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008289365210222589,
+      "loss": 5.9919,
+      "step": 1334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008287304204451772,
+      "loss": 5.2768,
+      "step": 1335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008285243198680956,
+      "loss": 5.8796,
+      "step": 1336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000828318219291014,
+      "loss": 5.9732,
+      "step": 1337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008281121187139324,
+      "loss": 5.586,
+      "step": 1338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008279060181368509,
+      "loss": 5.9313,
+      "step": 1339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008276999175597691,
+      "loss": 6.1477,
+      "step": 1340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008274938169826876,
+      "loss": 5.845,
+      "step": 1341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000827287716405606,
+      "loss": 5.6516,
+      "step": 1342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008270816158285244,
+      "loss": 5.8269,
+      "step": 1343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008268755152514427,
+      "loss": 6.1864,
+      "step": 1344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008266694146743611,
+      "loss": 6.3781,
+      "step": 1345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0008264633140972795,
+      "loss": 5.5599,
+      "step": 1346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008262572135201979,
+      "loss": 6.9528,
+      "step": 1347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008260511129431163,
+      "loss": 5.174,
+      "step": 1348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008258450123660346,
+      "loss": 6.4844,
+      "step": 1349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000825638911788953,
+      "loss": 5.0703,
+      "step": 1350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008254328112118714,
+      "loss": 5.8515,
+      "step": 1351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008252267106347898,
+      "loss": 5.9545,
+      "step": 1352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008250206100577082,
+      "loss": 5.9305,
+      "step": 1353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008248145094806265,
+      "loss": 6.0371,
+      "step": 1354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008246084089035449,
+      "loss": 5.4913,
+      "step": 1355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008244023083264633,
+      "loss": 5.9922,
+      "step": 1356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008241962077493817,
+      "loss": 5.3029,
+      "step": 1357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008239901071723,
+      "loss": 6.07,
+      "step": 1358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008237840065952184,
+      "loss": 5.7566,
+      "step": 1359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008235779060181368,
+      "loss": 5.9066,
+      "step": 1360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008233718054410553,
+      "loss": 5.5343,
+      "step": 1361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008231657048639737,
+      "loss": 5.5452,
+      "step": 1362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000822959604286892,
+      "loss": 5.3677,
+      "step": 1363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0008227535037098104,
+      "loss": 6.5452,
+      "step": 1364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008225474031327288,
+      "loss": 5.6282,
+      "step": 1365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008223413025556472,
+      "loss": 5.5065,
+      "step": 1366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008221352019785655,
+      "loss": 5.7555,
+      "step": 1367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008219291014014839,
+      "loss": 5.3325,
+      "step": 1368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008217230008244023,
+      "loss": 5.5145,
+      "step": 1369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008215169002473207,
+      "loss": 5.6403,
+      "step": 1370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008213107996702391,
+      "loss": 5.8506,
+      "step": 1371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008211046990931574,
+      "loss": 5.2348,
+      "step": 1372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008208985985160758,
+      "loss": 5.8213,
+      "step": 1373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008206924979389942,
+      "loss": 5.6963,
+      "step": 1374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008204863973619126,
+      "loss": 5.4453,
+      "step": 1375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008202802967848309,
+      "loss": 5.1995,
+      "step": 1376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008200741962077493,
+      "loss": 5.9269,
+      "step": 1377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008198680956306677,
+      "loss": 5.8069,
+      "step": 1378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008196619950535861,
+      "loss": 5.4961,
+      "step": 1379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008194558944765046,
+      "loss": 5.5949,
+      "step": 1380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008192497938994228,
+      "loss": 5.1319,
+      "step": 1381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0008190436933223413,
+      "loss": 5.5332,
+      "step": 1382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008188375927452597,
+      "loss": 5.3789,
+      "step": 1383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008186314921681781,
+      "loss": 6.3379,
+      "step": 1384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008184253915910965,
+      "loss": 5.6766,
+      "step": 1385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008182192910140148,
+      "loss": 5.5301,
+      "step": 1386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008180131904369332,
+      "loss": 5.7618,
+      "step": 1387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008178070898598516,
+      "loss": 5.6071,
+      "step": 1388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00081760098928277,
+      "loss": 5.3602,
+      "step": 1389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008173948887056883,
+      "loss": 6.3751,
+      "step": 1390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008171887881286068,
+      "loss": 5.3816,
+      "step": 1391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008169826875515252,
+      "loss": 5.8244,
+      "step": 1392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008167765869744436,
+      "loss": 5.3935,
+      "step": 1393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000816570486397362,
+      "loss": 5.7353,
+      "step": 1394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008163643858202803,
+      "loss": 6.2475,
+      "step": 1395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008161582852431987,
+      "loss": 5.3608,
+      "step": 1396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008159521846661171,
+      "loss": 5.6385,
+      "step": 1397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008157460840890355,
+      "loss": 5.2458,
+      "step": 1398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008155399835119538,
+      "loss": 5.42,
+      "step": 1399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0008153338829348723,
+      "loss": 5.4052,
+      "step": 1400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008151277823577907,
+      "loss": 5.4661,
+      "step": 1401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008149216817807091,
+      "loss": 6.0852,
+      "step": 1402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008147155812036275,
+      "loss": 5.4356,
+      "step": 1403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008145094806265458,
+      "loss": 6.6441,
+      "step": 1404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008143033800494642,
+      "loss": 5.5551,
+      "step": 1405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008140972794723826,
+      "loss": 6.7382,
+      "step": 1406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000813891178895301,
+      "loss": 5.6702,
+      "step": 1407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008136850783182194,
+      "loss": 5.6394,
+      "step": 1408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008134789777411377,
+      "loss": 5.6466,
+      "step": 1409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008132728771640561,
+      "loss": 5.7397,
+      "step": 1410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008130667765869745,
+      "loss": 5.6162,
+      "step": 1411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008128606760098929,
+      "loss": 5.2814,
+      "step": 1412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008126545754328112,
+      "loss": 5.7554,
+      "step": 1413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008124484748557296,
+      "loss": 6.1275,
+      "step": 1414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000812242374278648,
+      "loss": 5.4784,
+      "step": 1415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008120362737015664,
+      "loss": 5.8412,
+      "step": 1416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008118301731244848,
+      "loss": 5.6736,
+      "step": 1417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0008116240725474031,
+      "loss": 6.9356,
+      "step": 1418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008114179719703215,
+      "loss": 5.8975,
+      "step": 1419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00081121187139324,
+      "loss": 5.4791,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008110057708161584,
+      "loss": 5.7568,
+      "step": 1421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008107996702390767,
+      "loss": 5.8009,
+      "step": 1422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008105935696619951,
+      "loss": 5.4716,
+      "step": 1423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008103874690849135,
+      "loss": 5.7019,
+      "step": 1424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008101813685078319,
+      "loss": 6.5823,
+      "step": 1425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008099752679307503,
+      "loss": 5.652,
+      "step": 1426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008097691673536686,
+      "loss": 5.3077,
+      "step": 1427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000809563066776587,
+      "loss": 5.5405,
+      "step": 1428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008093569661995054,
+      "loss": 5.8158,
+      "step": 1429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008091508656224238,
+      "loss": 6.6466,
+      "step": 1430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008089447650453421,
+      "loss": 7.0253,
+      "step": 1431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008087386644682605,
+      "loss": 5.8179,
+      "step": 1432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008085325638911789,
+      "loss": 6.1565,
+      "step": 1433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008083264633140973,
+      "loss": 5.5547,
+      "step": 1434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0008081203627370157,
+      "loss": 5.6899,
+      "step": 1435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000807914262159934,
+      "loss": 5.4282,
+      "step": 1436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008077081615828524,
+      "loss": 5.6716,
+      "step": 1437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008075020610057708,
+      "loss": 5.458,
+      "step": 1438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008072959604286892,
+      "loss": 5.4726,
+      "step": 1439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008070898598516077,
+      "loss": 5.7897,
+      "step": 1440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000806883759274526,
+      "loss": 6.5274,
+      "step": 1441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008066776586974444,
+      "loss": 5.5945,
+      "step": 1442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008064715581203628,
+      "loss": 5.321,
+      "step": 1443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008062654575432812,
+      "loss": 5.4275,
+      "step": 1444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008060593569661995,
+      "loss": 5.4809,
+      "step": 1445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008058532563891179,
+      "loss": 5.6175,
+      "step": 1446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008056471558120363,
+      "loss": 5.5429,
+      "step": 1447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008054410552349547,
+      "loss": 5.3064,
+      "step": 1448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008052349546578731,
+      "loss": 6.3662,
+      "step": 1449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008050288540807914,
+      "loss": 5.4939,
+      "step": 1450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008048227535037098,
+      "loss": 5.7571,
+      "step": 1451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008046166529266282,
+      "loss": 5.6614,
+      "step": 1452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0008044105523495466,
+      "loss": 5.6784,
+      "step": 1453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008042044517724649,
+      "loss": 5.4909,
+      "step": 1454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008039983511953833,
+      "loss": 7.2386,
+      "step": 1455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008037922506183017,
+      "loss": 5.7242,
+      "step": 1456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008035861500412201,
+      "loss": 5.9933,
+      "step": 1457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008033800494641385,
+      "loss": 6.0223,
+      "step": 1458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008031739488870568,
+      "loss": 5.5806,
+      "step": 1459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008029678483099752,
+      "loss": 5.9806,
+      "step": 1460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008027617477328936,
+      "loss": 5.6577,
+      "step": 1461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008025556471558121,
+      "loss": 5.8956,
+      "step": 1462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008023495465787304,
+      "loss": 5.4332,
+      "step": 1463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008021434460016488,
+      "loss": 5.6937,
+      "step": 1464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008019373454245672,
+      "loss": 6.153,
+      "step": 1465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008017312448474856,
+      "loss": 5.7969,
+      "step": 1466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000801525144270404,
+      "loss": 5.5549,
+      "step": 1467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008013190436933223,
+      "loss": 5.6995,
+      "step": 1468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008011129431162407,
+      "loss": 5.7074,
+      "step": 1469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008009068425391591,
+      "loss": 5.4543,
+      "step": 1470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0008007007419620775,
+      "loss": 5.6145,
+      "step": 1471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0008004946413849959,
+      "loss": 5.2262,
+      "step": 1472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0008002885408079142,
+      "loss": 5.3823,
+      "step": 1473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0008000824402308326,
+      "loss": 5.3046,
+      "step": 1474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000799876339653751,
+      "loss": 5.9789,
+      "step": 1475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007996702390766694,
+      "loss": 6.2284,
+      "step": 1476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007994641384995877,
+      "loss": 5.7527,
+      "step": 1477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007992580379225061,
+      "loss": 5.2472,
+      "step": 1478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007990519373454245,
+      "loss": 5.4586,
+      "step": 1479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007988458367683429,
+      "loss": 5.2274,
+      "step": 1480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007986397361912613,
+      "loss": 5.8678,
+      "step": 1481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007984336356141796,
+      "loss": 5.907,
+      "step": 1482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000798227535037098,
+      "loss": 5.5314,
+      "step": 1483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007980214344600166,
+      "loss": 5.3772,
+      "step": 1484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000797815333882935,
+      "loss": 5.514,
+      "step": 1485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007976092333058533,
+      "loss": 5.8159,
+      "step": 1486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007974031327287717,
+      "loss": 5.65,
+      "step": 1487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007971970321516901,
+      "loss": 5.6132,
+      "step": 1488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0007969909315746085,
+      "loss": 5.9833,
+      "step": 1489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007967848309975269,
+      "loss": 5.2697,
+      "step": 1490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007965787304204452,
+      "loss": 5.8207,
+      "step": 1491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007963726298433636,
+      "loss": 6.1529,
+      "step": 1492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000796166529266282,
+      "loss": 4.8965,
+      "step": 1493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007959604286892004,
+      "loss": 6.0289,
+      "step": 1494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007957543281121187,
+      "loss": 5.2188,
+      "step": 1495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007955482275350371,
+      "loss": 7.7199,
+      "step": 1496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007953421269579555,
+      "loss": 5.8015,
+      "step": 1497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007951360263808739,
+      "loss": 5.683,
+      "step": 1498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007949299258037923,
+      "loss": 5.0551,
+      "step": 1499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007947238252267106,
+      "loss": 5.4896,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 5.801449775695801,
+      "eval_runtime": 1253.4482,
+      "eval_samples_per_second": 2.108,
+      "eval_steps_per_second": 0.264,
+      "eval_wer": 1.9549385164617215,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000794517724649629,
+      "loss": 5.5063,
+      "step": 1501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007943116240725475,
+      "loss": 5.8259,
+      "step": 1502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007941055234954659,
+      "loss": 5.966,
+      "step": 1503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007938994229183843,
+      "loss": 6.0704,
+      "step": 1504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007936933223413026,
+      "loss": 5.6497,
+      "step": 1505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000793487221764221,
+      "loss": 5.6551,
+      "step": 1506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0007932811211871394,
+      "loss": 5.3725,
+      "step": 1507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007930750206100578,
+      "loss": 5.6466,
+      "step": 1508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007928689200329761,
+      "loss": 5.5808,
+      "step": 1509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007926628194558945,
+      "loss": 5.6186,
+      "step": 1510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007924567188788129,
+      "loss": 5.3085,
+      "step": 1511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007922506183017313,
+      "loss": 5.5117,
+      "step": 1512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007920445177246497,
+      "loss": 5.1488,
+      "step": 1513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000791838417147568,
+      "loss": 5.2962,
+      "step": 1514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007916323165704864,
+      "loss": 5.5465,
+      "step": 1515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007914262159934048,
+      "loss": 6.092,
+      "step": 1516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007912201154163232,
+      "loss": 5.3069,
+      "step": 1517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007910140148392415,
+      "loss": 5.7516,
+      "step": 1518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007908079142621599,
+      "loss": 5.7588,
+      "step": 1519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007906018136850783,
+      "loss": 5.541,
+      "step": 1520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007903957131079967,
+      "loss": 5.8282,
+      "step": 1521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007901896125309152,
+      "loss": 5.7781,
+      "step": 1522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007899835119538335,
+      "loss": 6.2454,
+      "step": 1523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007897774113767519,
+      "loss": 5.4399,
+      "step": 1524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0007895713107996703,
+      "loss": 5.2182,
+      "step": 1525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007893652102225887,
+      "loss": 5.6154,
+      "step": 1526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007891591096455071,
+      "loss": 5.5759,
+      "step": 1527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007889530090684254,
+      "loss": 5.3504,
+      "step": 1528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007887469084913438,
+      "loss": 5.5403,
+      "step": 1529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007885408079142622,
+      "loss": 5.3765,
+      "step": 1530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007883347073371806,
+      "loss": 5.4516,
+      "step": 1531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007881286067600989,
+      "loss": 5.9716,
+      "step": 1532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007879225061830173,
+      "loss": 5.7073,
+      "step": 1533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007877164056059357,
+      "loss": 5.4937,
+      "step": 1534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007875103050288541,
+      "loss": 5.6473,
+      "step": 1535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007873042044517725,
+      "loss": 5.276,
+      "step": 1536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007870981038746908,
+      "loss": 5.8877,
+      "step": 1537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007868920032976092,
+      "loss": 5.6901,
+      "step": 1538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007866859027205276,
+      "loss": 5.5254,
+      "step": 1539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000786479802143446,
+      "loss": 5.7,
+      "step": 1540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007862737015663643,
+      "loss": 5.9088,
+      "step": 1541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007860676009892827,
+      "loss": 5.1651,
+      "step": 1542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0007858615004122012,
+      "loss": 5.7937,
+      "step": 1543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007856553998351196,
+      "loss": 6.0924,
+      "step": 1544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000785449299258038,
+      "loss": 6.2691,
+      "step": 1545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007852431986809563,
+      "loss": 5.2774,
+      "step": 1546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007850370981038747,
+      "loss": 5.6981,
+      "step": 1547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007848309975267931,
+      "loss": 5.5326,
+      "step": 1548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007846248969497115,
+      "loss": 5.8238,
+      "step": 1549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007844187963726298,
+      "loss": 5.1128,
+      "step": 1550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007842126957955482,
+      "loss": 5.8308,
+      "step": 1551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007840065952184666,
+      "loss": 5.7238,
+      "step": 1552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000783800494641385,
+      "loss": 5.4008,
+      "step": 1553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007835943940643034,
+      "loss": 5.2363,
+      "step": 1554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007833882934872217,
+      "loss": 5.6624,
+      "step": 1555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007831821929101401,
+      "loss": 6.0053,
+      "step": 1556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007829760923330585,
+      "loss": 5.6289,
+      "step": 1557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007827699917559769,
+      "loss": 5.7722,
+      "step": 1558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007825638911788953,
+      "loss": 5.636,
+      "step": 1559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0007823577906018136,
+      "loss": 5.7234,
+      "step": 1560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000782151690024732,
+      "loss": 5.7175,
+      "step": 1561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007819455894476504,
+      "loss": 5.9565,
+      "step": 1562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007817394888705689,
+      "loss": 5.6835,
+      "step": 1563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007815333882934872,
+      "loss": 5.3139,
+      "step": 1564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007813272877164056,
+      "loss": 5.5109,
+      "step": 1565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000781121187139324,
+      "loss": 5.6121,
+      "step": 1566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007809150865622424,
+      "loss": 5.2496,
+      "step": 1567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007807089859851608,
+      "loss": 5.6109,
+      "step": 1568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007805028854080791,
+      "loss": 5.8406,
+      "step": 1569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007802967848309975,
+      "loss": 5.6193,
+      "step": 1570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007800906842539159,
+      "loss": 5.4892,
+      "step": 1571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007798845836768343,
+      "loss": 5.6991,
+      "step": 1572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007796784830997526,
+      "loss": 5.5619,
+      "step": 1573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000779472382522671,
+      "loss": 5.1656,
+      "step": 1574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007792662819455894,
+      "loss": 5.5771,
+      "step": 1575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007790601813685078,
+      "loss": 5.6983,
+      "step": 1576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007788540807914263,
+      "loss": 6.0226,
+      "step": 1577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0007786479802143446,
+      "loss": 5.7092,
+      "step": 1578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000778441879637263,
+      "loss": 5.9255,
+      "step": 1579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007782357790601814,
+      "loss": 5.8465,
+      "step": 1580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007780296784830998,
+      "loss": 5.6712,
+      "step": 1581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007778235779060181,
+      "loss": 5.5361,
+      "step": 1582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007776174773289366,
+      "loss": 5.6805,
+      "step": 1583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000777411376751855,
+      "loss": 5.3949,
+      "step": 1584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007772052761747734,
+      "loss": 5.4973,
+      "step": 1585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007769991755976918,
+      "loss": 5.2959,
+      "step": 1586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007767930750206101,
+      "loss": 5.8502,
+      "step": 1587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007765869744435285,
+      "loss": 5.5001,
+      "step": 1588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007763808738664469,
+      "loss": 5.5013,
+      "step": 1589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007761747732893653,
+      "loss": 5.325,
+      "step": 1590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007759686727122837,
+      "loss": 5.3286,
+      "step": 1591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000775762572135202,
+      "loss": 5.3916,
+      "step": 1592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007755564715581204,
+      "loss": 5.3245,
+      "step": 1593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007753503709810388,
+      "loss": 5.2796,
+      "step": 1594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007751442704039572,
+      "loss": 5.3226,
+      "step": 1595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0007749381698268755,
+      "loss": 5.467,
+      "step": 1596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007747320692497939,
+      "loss": 5.6445,
+      "step": 1597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007745259686727123,
+      "loss": 5.4956,
+      "step": 1598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007743198680956307,
+      "loss": 6.3905,
+      "step": 1599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007741137675185491,
+      "loss": 5.5943,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007739076669414674,
+      "loss": 5.7323,
+      "step": 1601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007737015663643858,
+      "loss": 5.2754,
+      "step": 1602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007734954657873043,
+      "loss": 6.3494,
+      "step": 1603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007732893652102227,
+      "loss": 5.5458,
+      "step": 1604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000773083264633141,
+      "loss": 5.131,
+      "step": 1605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007728771640560594,
+      "loss": 5.3759,
+      "step": 1606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007726710634789778,
+      "loss": 5.5309,
+      "step": 1607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007724649629018962,
+      "loss": 6.3347,
+      "step": 1608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007722588623248146,
+      "loss": 5.6963,
+      "step": 1609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007720527617477329,
+      "loss": 5.7869,
+      "step": 1610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007718466611706513,
+      "loss": 5.5897,
+      "step": 1611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007716405605935697,
+      "loss": 5.6985,
+      "step": 1612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007714344600164881,
+      "loss": 6.2111,
+      "step": 1613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0007712283594394065,
+      "loss": 5.2933,
+      "step": 1614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007710222588623248,
+      "loss": 5.3394,
+      "step": 1615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007708161582852432,
+      "loss": 5.3486,
+      "step": 1616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007706100577081616,
+      "loss": 5.5827,
+      "step": 1617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00077040395713108,
+      "loss": 5.5502,
+      "step": 1618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007701978565539983,
+      "loss": 5.3902,
+      "step": 1619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007699917559769167,
+      "loss": 5.3336,
+      "step": 1620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007697856553998351,
+      "loss": 5.7054,
+      "step": 1621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007695795548227535,
+      "loss": 5.1106,
+      "step": 1622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000769373454245672,
+      "loss": 5.423,
+      "step": 1623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007691673536685903,
+      "loss": 5.2948,
+      "step": 1624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007689612530915087,
+      "loss": 5.5325,
+      "step": 1625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007687551525144271,
+      "loss": 5.7764,
+      "step": 1626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007685490519373455,
+      "loss": 5.6459,
+      "step": 1627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007683429513602638,
+      "loss": 5.2827,
+      "step": 1628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007681368507831822,
+      "loss": 5.4788,
+      "step": 1629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007679307502061006,
+      "loss": 5.5417,
+      "step": 1630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000767724649629019,
+      "loss": 5.3799,
+      "step": 1631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0007675185490519374,
+      "loss": 5.9044,
+      "step": 1632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007673124484748557,
+      "loss": 5.9922,
+      "step": 1633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007671063478977741,
+      "loss": 5.3713,
+      "step": 1634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007669002473206925,
+      "loss": 5.4643,
+      "step": 1635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007666941467436109,
+      "loss": 5.786,
+      "step": 1636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007664880461665292,
+      "loss": 5.1989,
+      "step": 1637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007662819455894476,
+      "loss": 6.0217,
+      "step": 1638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000766075845012366,
+      "loss": 5.1845,
+      "step": 1639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007658697444352844,
+      "loss": 5.7345,
+      "step": 1640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007656636438582028,
+      "loss": 5.7359,
+      "step": 1641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007654575432811211,
+      "loss": 5.1966,
+      "step": 1642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007652514427040395,
+      "loss": 5.409,
+      "step": 1643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000765045342126958,
+      "loss": 5.4155,
+      "step": 1644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007648392415498764,
+      "loss": 6.0461,
+      "step": 1645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007646331409727948,
+      "loss": 5.8304,
+      "step": 1646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007644270403957131,
+      "loss": 5.2406,
+      "step": 1647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007642209398186315,
+      "loss": 6.1266,
+      "step": 1648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007640148392415499,
+      "loss": 5.7275,
+      "step": 1649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0007638087386644683,
+      "loss": 5.1881,
+      "step": 1650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007636026380873866,
+      "loss": 6.7051,
+      "step": 1651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000763396537510305,
+      "loss": 5.4683,
+      "step": 1652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007631904369332234,
+      "loss": 5.7139,
+      "step": 1653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007629843363561418,
+      "loss": 5.4164,
+      "step": 1654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007627782357790602,
+      "loss": 5.2561,
+      "step": 1655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007625721352019785,
+      "loss": 5.3964,
+      "step": 1656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007623660346248969,
+      "loss": 5.9503,
+      "step": 1657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007621599340478153,
+      "loss": 5.6585,
+      "step": 1658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007619538334707337,
+      "loss": 5.5381,
+      "step": 1659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000761747732893652,
+      "loss": 5.3597,
+      "step": 1660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007615416323165704,
+      "loss": 5.4913,
+      "step": 1661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007613355317394888,
+      "loss": 5.2972,
+      "step": 1662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007611294311624072,
+      "loss": 5.1786,
+      "step": 1663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007609233305853257,
+      "loss": 5.5845,
+      "step": 1664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000760717230008244,
+      "loss": 5.2914,
+      "step": 1665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007605111294311624,
+      "loss": 5.5602,
+      "step": 1666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007603050288540808,
+      "loss": 5.4863,
+      "step": 1667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0007600989282769992,
+      "loss": 5.7145,
+      "step": 1668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007598928276999175,
+      "loss": 5.1099,
+      "step": 1669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007596867271228359,
+      "loss": 5.7397,
+      "step": 1670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007594806265457544,
+      "loss": 5.7404,
+      "step": 1671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007592745259686728,
+      "loss": 5.9187,
+      "step": 1672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007590684253915912,
+      "loss": 5.4125,
+      "step": 1673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007588623248145095,
+      "loss": 5.4871,
+      "step": 1674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007586562242374279,
+      "loss": 5.9435,
+      "step": 1675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007584501236603463,
+      "loss": 5.2349,
+      "step": 1676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007582440230832647,
+      "loss": 5.1204,
+      "step": 1677
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007580379225061831,
+      "loss": 5.2944,
+      "step": 1678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007578318219291014,
+      "loss": 5.5846,
+      "step": 1679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007576257213520198,
+      "loss": 5.9795,
+      "step": 1680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007574196207749382,
+      "loss": 5.0281,
+      "step": 1681
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007572135201978566,
+      "loss": 5.2559,
+      "step": 1682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000757007419620775,
+      "loss": 5.442,
+      "step": 1683
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007568013190436934,
+      "loss": 5.456,
+      "step": 1684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0007565952184666118,
+      "loss": 5.6921,
+      "step": 1685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007563891178895302,
+      "loss": 5.364,
+      "step": 1686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007561830173124486,
+      "loss": 5.6626,
+      "step": 1687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007559769167353669,
+      "loss": 5.718,
+      "step": 1688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007557708161582853,
+      "loss": 5.5039,
+      "step": 1689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007555647155812037,
+      "loss": 5.0712,
+      "step": 1690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007553586150041221,
+      "loss": 5.9504,
+      "step": 1691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007551525144270404,
+      "loss": 5.792,
+      "step": 1692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007549464138499588,
+      "loss": 5.59,
+      "step": 1693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007547403132728772,
+      "loss": 5.72,
+      "step": 1694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007545342126957956,
+      "loss": 5.5693,
+      "step": 1695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000754328112118714,
+      "loss": 8.4865,
+      "step": 1696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007541220115416323,
+      "loss": 5.2336,
+      "step": 1697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007539159109645507,
+      "loss": 5.765,
+      "step": 1698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007537098103874691,
+      "loss": 5.6956,
+      "step": 1699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007535037098103875,
+      "loss": 5.9469,
+      "step": 1700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007532976092333058,
+      "loss": 6.342,
+      "step": 1701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007530915086562242,
+      "loss": 5.7973,
+      "step": 1702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0007528854080791426,
+      "loss": 5.8899,
+      "step": 1703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000752679307502061,
+      "loss": 5.8766,
+      "step": 1704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007524732069249795,
+      "loss": 5.6317,
+      "step": 1705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007522671063478978,
+      "loss": 5.3092,
+      "step": 1706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007520610057708162,
+      "loss": 5.3854,
+      "step": 1707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007518549051937346,
+      "loss": 5.71,
+      "step": 1708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000751648804616653,
+      "loss": 5.9464,
+      "step": 1709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007514427040395714,
+      "loss": 5.9005,
+      "step": 1710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007512366034624897,
+      "loss": 5.2014,
+      "step": 1711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007510305028854081,
+      "loss": 5.6068,
+      "step": 1712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007508244023083265,
+      "loss": 6.0353,
+      "step": 1713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007506183017312449,
+      "loss": 5.5371,
+      "step": 1714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007504122011541632,
+      "loss": 5.341,
+      "step": 1715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007502061005770816,
+      "loss": 6.2251,
+      "step": 1716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00075,
+      "loss": 6.5625,
+      "step": 1717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007497938994229184,
+      "loss": 6.3635,
+      "step": 1718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007495877988458368,
+      "loss": 6.022,
+      "step": 1719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007493816982687551,
+      "loss": 5.5182,
+      "step": 1720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0007491755976916735,
+      "loss": 5.417,
+      "step": 1721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007489694971145919,
+      "loss": 5.8588,
+      "step": 1722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007487633965375103,
+      "loss": 5.9432,
+      "step": 1723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007485572959604286,
+      "loss": 5.4998,
+      "step": 1724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000748351195383347,
+      "loss": 5.4777,
+      "step": 1725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007481450948062655,
+      "loss": 5.4144,
+      "step": 1726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007479389942291839,
+      "loss": 5.5469,
+      "step": 1727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007477328936521023,
+      "loss": 5.7742,
+      "step": 1728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007475267930750206,
+      "loss": 6.2265,
+      "step": 1729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000747320692497939,
+      "loss": 6.1309,
+      "step": 1730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007471145919208574,
+      "loss": 5.0155,
+      "step": 1731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007469084913437758,
+      "loss": 5.3373,
+      "step": 1732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007467023907666942,
+      "loss": 5.3111,
+      "step": 1733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007464962901896125,
+      "loss": 5.5686,
+      "step": 1734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007462901896125309,
+      "loss": 6.0966,
+      "step": 1735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007460840890354493,
+      "loss": 6.1853,
+      "step": 1736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007458779884583677,
+      "loss": 5.0348,
+      "step": 1737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000745671887881286,
+      "loss": 6.2651,
+      "step": 1738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0007454657873042044,
+      "loss": 6.321,
+      "step": 1739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007452596867271228,
+      "loss": 5.7426,
+      "step": 1740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007450535861500412,
+      "loss": 7.2042,
+      "step": 1741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007448474855729596,
+      "loss": 5.422,
+      "step": 1742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007446413849958779,
+      "loss": 5.6667,
+      "step": 1743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007444352844187963,
+      "loss": 5.6673,
+      "step": 1744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007442291838417148,
+      "loss": 5.7163,
+      "step": 1745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007440230832646332,
+      "loss": 5.3528,
+      "step": 1746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007438169826875515,
+      "loss": 5.5586,
+      "step": 1747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007436108821104699,
+      "loss": 5.0461,
+      "step": 1748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007434047815333883,
+      "loss": 5.2546,
+      "step": 1749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007431986809563067,
+      "loss": 6.7615,
+      "step": 1750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007429925803792251,
+      "loss": 5.292,
+      "step": 1751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007427864798021434,
+      "loss": 5.2499,
+      "step": 1752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007425803792250618,
+      "loss": 5.7262,
+      "step": 1753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007423742786479802,
+      "loss": 6.2522,
+      "step": 1754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007421681780708986,
+      "loss": 5.602,
+      "step": 1755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007419620774938169,
+      "loss": 5.4795,
+      "step": 1756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0007417559769167353,
+      "loss": 5.263,
+      "step": 1757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007415498763396537,
+      "loss": 5.3496,
+      "step": 1758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007413437757625721,
+      "loss": 5.8711,
+      "step": 1759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007411376751854905,
+      "loss": 5.3811,
+      "step": 1760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007409315746084088,
+      "loss": 5.7148,
+      "step": 1761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007407254740313272,
+      "loss": 5.2924,
+      "step": 1762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007405193734542456,
+      "loss": 5.5371,
+      "step": 1763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007403132728771642,
+      "loss": 5.6456,
+      "step": 1764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007401071723000826,
+      "loss": 5.3024,
+      "step": 1765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007399010717230009,
+      "loss": 5.5201,
+      "step": 1766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007396949711459193,
+      "loss": 5.4668,
+      "step": 1767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007394888705688377,
+      "loss": 5.2057,
+      "step": 1768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007392827699917561,
+      "loss": 5.297,
+      "step": 1769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007390766694146744,
+      "loss": 5.2295,
+      "step": 1770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007388705688375928,
+      "loss": 5.5304,
+      "step": 1771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007386644682605112,
+      "loss": 5.2296,
+      "step": 1772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007384583676834296,
+      "loss": 5.3524,
+      "step": 1773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000738252267106348,
+      "loss": 5.3564,
+      "step": 1774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0007380461665292663,
+      "loss": 5.1979,
+      "step": 1775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007378400659521847,
+      "loss": 5.2965,
+      "step": 1776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007376339653751031,
+      "loss": 6.6263,
+      "step": 1777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007374278647980215,
+      "loss": 5.4154,
+      "step": 1778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007372217642209398,
+      "loss": 7.0764,
+      "step": 1779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007370156636438582,
+      "loss": 5.3152,
+      "step": 1780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007368095630667766,
+      "loss": 5.9469,
+      "step": 1781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000736603462489695,
+      "loss": 5.7224,
+      "step": 1782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007363973619126134,
+      "loss": 6.7119,
+      "step": 1783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007361912613355317,
+      "loss": 4.5925,
+      "step": 1784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007359851607584502,
+      "loss": 6.4766,
+      "step": 1785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007357790601813686,
+      "loss": 5.8479,
+      "step": 1786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000735572959604287,
+      "loss": 5.1044,
+      "step": 1787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007353668590272053,
+      "loss": 5.338,
+      "step": 1788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007351607584501237,
+      "loss": 5.4172,
+      "step": 1789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007349546578730421,
+      "loss": 5.8801,
+      "step": 1790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007347485572959605,
+      "loss": 5.0918,
+      "step": 1791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0007345424567188789,
+      "loss": 5.0638,
+      "step": 1792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007343363561417972,
+      "loss": 5.5677,
+      "step": 1793
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007341302555647156,
+      "loss": 5.1254,
+      "step": 1794
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000733924154987634,
+      "loss": 5.3448,
+      "step": 1795
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007337180544105524,
+      "loss": 5.3097,
+      "step": 1796
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007335119538334708,
+      "loss": 4.8294,
+      "step": 1797
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007333058532563891,
+      "loss": 5.4026,
+      "step": 1798
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007330997526793075,
+      "loss": 5.3028,
+      "step": 1799
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007328936521022259,
+      "loss": 5.4134,
+      "step": 1800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007326875515251443,
+      "loss": 5.1644,
+      "step": 1801
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007324814509480626,
+      "loss": 5.0567,
+      "step": 1802
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000732275350370981,
+      "loss": 5.2201,
+      "step": 1803
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007320692497938994,
+      "loss": 5.4445,
+      "step": 1804
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007318631492168179,
+      "loss": 4.9351,
+      "step": 1805
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007316570486397363,
+      "loss": 5.2178,
+      "step": 1806
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007314509480626546,
+      "loss": 4.8675,
+      "step": 1807
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000731244847485573,
+      "loss": 5.4179,
+      "step": 1808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007310387469084914,
+      "loss": 5.1806,
+      "step": 1809
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0007308326463314098,
+      "loss": 4.9518,
+      "step": 1810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007306265457543281,
+      "loss": 5.2499,
+      "step": 1811
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007304204451772465,
+      "loss": 6.2926,
+      "step": 1812
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007302143446001649,
+      "loss": 5.117,
+      "step": 1813
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007300082440230833,
+      "loss": 4.6867,
+      "step": 1814
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007298021434460017,
+      "loss": 5.2491,
+      "step": 1815
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00072959604286892,
+      "loss": 6.3636,
+      "step": 1816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007293899422918384,
+      "loss": 5.2329,
+      "step": 1817
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007291838417147568,
+      "loss": 5.211,
+      "step": 1818
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007289777411376752,
+      "loss": 5.4492,
+      "step": 1819
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007287716405605935,
+      "loss": 5.1223,
+      "step": 1820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007285655399835119,
+      "loss": 5.2691,
+      "step": 1821
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007283594394064303,
+      "loss": 5.1526,
+      "step": 1822
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007281533388293487,
+      "loss": 5.0936,
+      "step": 1823
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007279472382522671,
+      "loss": 5.6415,
+      "step": 1824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007277411376751854,
+      "loss": 4.9492,
+      "step": 1825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007275350370981039,
+      "loss": 5.463,
+      "step": 1826
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007273289365210223,
+      "loss": 5.0917,
+      "step": 1827
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0007271228359439407,
+      "loss": 5.2381,
+      "step": 1828
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007269167353668591,
+      "loss": 5.1342,
+      "step": 1829
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007267106347897774,
+      "loss": 5.5174,
+      "step": 1830
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007265045342126958,
+      "loss": 4.637,
+      "step": 1831
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007262984336356142,
+      "loss": 5.1934,
+      "step": 1832
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007260923330585326,
+      "loss": 7.3791,
+      "step": 1833
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007258862324814509,
+      "loss": 5.604,
+      "step": 1834
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007256801319043693,
+      "loss": 5.9677,
+      "step": 1835
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007254740313272877,
+      "loss": 6.0263,
+      "step": 1836
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007252679307502061,
+      "loss": 5.0087,
+      "step": 1837
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007250618301731245,
+      "loss": 5.3671,
+      "step": 1838
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007248557295960428,
+      "loss": 5.1236,
+      "step": 1839
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007246496290189612,
+      "loss": 5.1798,
+      "step": 1840
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007244435284418796,
+      "loss": 5.3328,
+      "step": 1841
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000724237427864798,
+      "loss": 5.1277,
+      "step": 1842
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007240313272877163,
+      "loss": 5.3592,
+      "step": 1843
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007238252267106347,
+      "loss": 5.8466,
+      "step": 1844
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007236191261335531,
+      "loss": 5.6739,
+      "step": 1845
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0007234130255564716,
+      "loss": 5.1129,
+      "step": 1846
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00072320692497939,
+      "loss": 6.1491,
+      "step": 1847
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007230008244023083,
+      "loss": 5.0162,
+      "step": 1848
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007227947238252267,
+      "loss": 5.1657,
+      "step": 1849
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007225886232481451,
+      "loss": 5.2096,
+      "step": 1850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007223825226710635,
+      "loss": 5.0301,
+      "step": 1851
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007221764220939819,
+      "loss": 5.6318,
+      "step": 1852
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007219703215169002,
+      "loss": 4.8731,
+      "step": 1853
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007217642209398186,
+      "loss": 5.2462,
+      "step": 1854
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000721558120362737,
+      "loss": 5.1238,
+      "step": 1855
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007213520197856554,
+      "loss": 5.0269,
+      "step": 1856
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007211459192085737,
+      "loss": 5.2101,
+      "step": 1857
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007209398186314922,
+      "loss": 5.0547,
+      "step": 1858
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007207337180544106,
+      "loss": 4.9697,
+      "step": 1859
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000720527617477329,
+      "loss": 5.0644,
+      "step": 1860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007203215169002474,
+      "loss": 4.8406,
+      "step": 1861
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007201154163231657,
+      "loss": 5.1595,
+      "step": 1862
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007199093157460841,
+      "loss": 4.9878,
+      "step": 1863
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0007197032151690025,
+      "loss": 5.3345,
+      "step": 1864
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000719497114591921,
+      "loss": 5.3522,
+      "step": 1865
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007192910140148393,
+      "loss": 4.6227,
+      "step": 1866
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007190849134377577,
+      "loss": 5.1555,
+      "step": 1867
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007188788128606761,
+      "loss": 4.9984,
+      "step": 1868
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007186727122835945,
+      "loss": 4.9427,
+      "step": 1869
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007184666117065129,
+      "loss": 5.6602,
+      "step": 1870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007182605111294312,
+      "loss": 5.1464,
+      "step": 1871
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007180544105523496,
+      "loss": 5.1839,
+      "step": 1872
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000717848309975268,
+      "loss": 4.8411,
+      "step": 1873
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007176422093981864,
+      "loss": 4.9992,
+      "step": 1874
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007174361088211047,
+      "loss": 5.2425,
+      "step": 1875
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007172300082440231,
+      "loss": 5.0372,
+      "step": 1876
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007170239076669415,
+      "loss": 5.1477,
+      "step": 1877
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007168178070898599,
+      "loss": 5.1354,
+      "step": 1878
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007166117065127783,
+      "loss": 5.3658,
+      "step": 1879
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007164056059356966,
+      "loss": 4.571,
+      "step": 1880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000716199505358615,
+      "loss": 4.7733,
+      "step": 1881
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0007159934047815334,
+      "loss": 5.543,
+      "step": 1882
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007157873042044518,
+      "loss": 4.4211,
+      "step": 1883
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007155812036273702,
+      "loss": 5.0537,
+      "step": 1884
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007153751030502885,
+      "loss": 4.9106,
+      "step": 1885
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000715169002473207,
+      "loss": 4.9786,
+      "step": 1886
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007149629018961254,
+      "loss": 5.4864,
+      "step": 1887
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007147568013190438,
+      "loss": 5.3496,
+      "step": 1888
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007145507007419621,
+      "loss": 4.9686,
+      "step": 1889
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007143446001648805,
+      "loss": 5.6113,
+      "step": 1890
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007141384995877989,
+      "loss": 5.0476,
+      "step": 1891
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007139323990107173,
+      "loss": 5.2357,
+      "step": 1892
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007137262984336357,
+      "loss": 5.2824,
+      "step": 1893
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.000713520197856554,
+      "loss": 4.8903,
+      "step": 1894
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007133140972794724,
+      "loss": 5.2474,
+      "step": 1895
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007131079967023908,
+      "loss": 5.1134,
+      "step": 1896
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007129018961253092,
+      "loss": 5.0281,
+      "step": 1897
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007126957955482275,
+      "loss": 5.3824,
+      "step": 1898
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0007124896949711459,
+      "loss": 5.2119,
+      "step": 1899
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007122835943940643,
+      "loss": 5.0685,
+      "step": 1900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007120774938169827,
+      "loss": 5.4825,
+      "step": 1901
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007118713932399011,
+      "loss": 4.9477,
+      "step": 1902
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007116652926628194,
+      "loss": 5.0455,
+      "step": 1903
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007114591920857378,
+      "loss": 5.3881,
+      "step": 1904
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007112530915086562,
+      "loss": 5.3305,
+      "step": 1905
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007110469909315747,
+      "loss": 5.8965,
+      "step": 1906
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000710840890354493,
+      "loss": 5.928,
+      "step": 1907
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007106347897774114,
+      "loss": 5.659,
+      "step": 1908
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007104286892003298,
+      "loss": 5.5795,
+      "step": 1909
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007102225886232482,
+      "loss": 5.2139,
+      "step": 1910
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007100164880461666,
+      "loss": 5.3317,
+      "step": 1911
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007098103874690849,
+      "loss": 5.3324,
+      "step": 1912
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007096042868920033,
+      "loss": 5.7317,
+      "step": 1913
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007093981863149217,
+      "loss": 5.3886,
+      "step": 1914
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007091920857378401,
+      "loss": 5.3804,
+      "step": 1915
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007089859851607585,
+      "loss": 5.1271,
+      "step": 1916
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0007087798845836768,
+      "loss": 5.1195,
+      "step": 1917
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007085737840065952,
+      "loss": 5.3734,
+      "step": 1918
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007083676834295136,
+      "loss": 5.0709,
+      "step": 1919
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000708161582852432,
+      "loss": 4.8298,
+      "step": 1920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007079554822753503,
+      "loss": 4.9467,
+      "step": 1921
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007077493816982687,
+      "loss": 5.7142,
+      "step": 1922
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007075432811211871,
+      "loss": 5.3252,
+      "step": 1923
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007073371805441055,
+      "loss": 5.1218,
+      "step": 1924
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007071310799670239,
+      "loss": 5.5352,
+      "step": 1925
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007069249793899422,
+      "loss": 4.9601,
+      "step": 1926
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007067188788128607,
+      "loss": 4.9484,
+      "step": 1927
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007065127782357791,
+      "loss": 5.1715,
+      "step": 1928
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007063066776586975,
+      "loss": 4.8564,
+      "step": 1929
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007061005770816158,
+      "loss": 4.5674,
+      "step": 1930
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007058944765045342,
+      "loss": 4.4388,
+      "step": 1931
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007056883759274526,
+      "loss": 7.3062,
+      "step": 1932
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000705482275350371,
+      "loss": 4.5762,
+      "step": 1933
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007052761747732894,
+      "loss": 4.6824,
+      "step": 1934
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0007050700741962077,
+      "loss": 5.3067,
+      "step": 1935
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007048639736191261,
+      "loss": 5.0733,
+      "step": 1936
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007046578730420445,
+      "loss": 5.2803,
+      "step": 1937
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007044517724649629,
+      "loss": 6.3454,
+      "step": 1938
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007042456718878813,
+      "loss": 5.1945,
+      "step": 1939
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007040395713107996,
+      "loss": 5.721,
+      "step": 1940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000703833470733718,
+      "loss": 4.9562,
+      "step": 1941
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007036273701566364,
+      "loss": 5.4393,
+      "step": 1942
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007034212695795548,
+      "loss": 5.6645,
+      "step": 1943
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007032151690024731,
+      "loss": 5.3318,
+      "step": 1944
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007030090684253915,
+      "loss": 5.1666,
+      "step": 1945
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007028029678483099,
+      "loss": 5.0721,
+      "step": 1946
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007025968672712284,
+      "loss": 5.9535,
+      "step": 1947
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007023907666941468,
+      "loss": 5.2234,
+      "step": 1948
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007021846661170651,
+      "loss": 5.3689,
+      "step": 1949
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007019785655399835,
+      "loss": 4.9733,
+      "step": 1950
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000701772464962902,
+      "loss": 5.448,
+      "step": 1951
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007015663643858204,
+      "loss": 4.7595,
+      "step": 1952
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0007013602638087387,
+      "loss": 5.7278,
+      "step": 1953
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0007011541632316571,
+      "loss": 5.0396,
+      "step": 1954
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0007009480626545755,
+      "loss": 5.2735,
+      "step": 1955
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0007007419620774939,
+      "loss": 5.2534,
+      "step": 1956
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0007005358615004123,
+      "loss": 5.2661,
+      "step": 1957
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0007003297609233306,
+      "loss": 5.1997,
+      "step": 1958
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000700123660346249,
+      "loss": 5.4179,
+      "step": 1959
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006999175597691674,
+      "loss": 4.8651,
+      "step": 1960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006997114591920858,
+      "loss": 5.2277,
+      "step": 1961
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006995053586150041,
+      "loss": 4.7882,
+      "step": 1962
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006992992580379225,
+      "loss": 5.0706,
+      "step": 1963
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006990931574608409,
+      "loss": 5.5236,
+      "step": 1964
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006988870568837593,
+      "loss": 5.5512,
+      "step": 1965
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006986809563066778,
+      "loss": 4.8478,
+      "step": 1966
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000698474855729596,
+      "loss": 4.9853,
+      "step": 1967
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006982687551525145,
+      "loss": 4.6403,
+      "step": 1968
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006980626545754329,
+      "loss": 5.3232,
+      "step": 1969
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006978565539983513,
+      "loss": 5.0923,
+      "step": 1970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0006976504534212697,
+      "loss": 5.7597,
+      "step": 1971
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000697444352844188,
+      "loss": 5.0139,
+      "step": 1972
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006972382522671064,
+      "loss": 4.9749,
+      "step": 1973
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006970321516900248,
+      "loss": 5.3646,
+      "step": 1974
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006968260511129432,
+      "loss": 5.5755,
+      "step": 1975
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006966199505358615,
+      "loss": 4.8937,
+      "step": 1976
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006964138499587799,
+      "loss": 5.3579,
+      "step": 1977
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006962077493816983,
+      "loss": 4.787,
+      "step": 1978
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006960016488046167,
+      "loss": 4.5655,
+      "step": 1979
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006957955482275351,
+      "loss": 5.7754,
+      "step": 1980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006955894476504534,
+      "loss": 5.6494,
+      "step": 1981
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006953833470733718,
+      "loss": 5.316,
+      "step": 1982
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006951772464962902,
+      "loss": 5.1819,
+      "step": 1983
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006949711459192086,
+      "loss": 4.5101,
+      "step": 1984
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006947650453421269,
+      "loss": 5.0929,
+      "step": 1985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006945589447650453,
+      "loss": 5.4557,
+      "step": 1986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006943528441879638,
+      "loss": 5.0741,
+      "step": 1987
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006941467436108822,
+      "loss": 5.5646,
+      "step": 1988
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0006939406430338006,
+      "loss": 5.1955,
+      "step": 1989
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006937345424567189,
+      "loss": 5.1484,
+      "step": 1990
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006935284418796373,
+      "loss": 5.336,
+      "step": 1991
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006933223413025557,
+      "loss": 4.9777,
+      "step": 1992
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006931162407254741,
+      "loss": 5.1013,
+      "step": 1993
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006929101401483924,
+      "loss": 5.2018,
+      "step": 1994
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006927040395713108,
+      "loss": 5.2258,
+      "step": 1995
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006924979389942292,
+      "loss": 5.6301,
+      "step": 1996
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006922918384171476,
+      "loss": 5.1471,
+      "step": 1997
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000692085737840066,
+      "loss": 5.6075,
+      "step": 1998
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006918796372629843,
+      "loss": 5.2687,
+      "step": 1999
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006916735366859027,
+      "loss": 4.9941,
+      "step": 2000
+    },
+    {
+      "epoch": 1.12,
+      "eval_loss": 5.711738586425781,
+      "eval_runtime": 1249.2821,
+      "eval_samples_per_second": 2.115,
+      "eval_steps_per_second": 0.265,
+      "eval_wer": 1.9887742959143198,
+      "step": 2000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006914674361088211,
+      "loss": 5.2656,
+      "step": 2001
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006912613355317395,
+      "loss": 4.8914,
+      "step": 2002
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006910552349546579,
+      "loss": 5.031,
+      "step": 2003
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006908491343775762,
+      "loss": 5.1447,
+      "step": 2004
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006906430338004946,
+      "loss": 4.7917,
+      "step": 2005
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000690436933223413,
+      "loss": 4.7943,
+      "step": 2006
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0006902308326463315,
+      "loss": 4.922,
+      "step": 2007
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006900247320692497,
+      "loss": 5.0741,
+      "step": 2008
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006898186314921682,
+      "loss": 4.9802,
+      "step": 2009
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006896125309150866,
+      "loss": 4.8857,
+      "step": 2010
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000689406430338005,
+      "loss": 4.8877,
+      "step": 2011
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006892003297609234,
+      "loss": 5.1555,
+      "step": 2012
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006889942291838417,
+      "loss": 5.4532,
+      "step": 2013
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006887881286067601,
+      "loss": 5.6917,
+      "step": 2014
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006885820280296785,
+      "loss": 5.0677,
+      "step": 2015
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006883759274525969,
+      "loss": 4.9572,
+      "step": 2016
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006881698268755152,
+      "loss": 5.2568,
+      "step": 2017
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006879637262984336,
+      "loss": 5.1832,
+      "step": 2018
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000687757625721352,
+      "loss": 4.9613,
+      "step": 2019
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006875515251442704,
+      "loss": 5.082,
+      "step": 2020
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006873454245671888,
+      "loss": 5.3497,
+      "step": 2021
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006871393239901071,
+      "loss": 4.8512,
+      "step": 2022
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006869332234130255,
+      "loss": 4.7783,
+      "step": 2023
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0006867271228359439,
+      "loss": 5.834,
+      "step": 2024
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006865210222588623,
+      "loss": 5.4813,
+      "step": 2025
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006863149216817806,
+      "loss": 5.1958,
+      "step": 2026
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000686108821104699,
+      "loss": 5.1472,
+      "step": 2027
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006859027205276174,
+      "loss": 4.9839,
+      "step": 2028
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006856966199505359,
+      "loss": 5.303,
+      "step": 2029
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006854905193734543,
+      "loss": 5.3901,
+      "step": 2030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006852844187963726,
+      "loss": 7.2275,
+      "step": 2031
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000685078318219291,
+      "loss": 4.5321,
+      "step": 2032
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006848722176422094,
+      "loss": 6.3331,
+      "step": 2033
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006846661170651278,
+      "loss": 5.2055,
+      "step": 2034
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006844600164880462,
+      "loss": 5.192,
+      "step": 2035
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006842539159109645,
+      "loss": 5.2922,
+      "step": 2036
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006840478153338829,
+      "loss": 5.3391,
+      "step": 2037
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006838417147568013,
+      "loss": 5.2896,
+      "step": 2038
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006836356141797197,
+      "loss": 5.1837,
+      "step": 2039
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000683429513602638,
+      "loss": 5.9111,
+      "step": 2040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006832234130255564,
+      "loss": 4.8318,
+      "step": 2041
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0006830173124484748,
+      "loss": 5.4362,
+      "step": 2042
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006828112118713932,
+      "loss": 5.2742,
+      "step": 2043
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006826051112943117,
+      "loss": 4.8643,
+      "step": 2044
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00068239901071723,
+      "loss": 5.6978,
+      "step": 2045
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006821929101401484,
+      "loss": 5.3586,
+      "step": 2046
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006819868095630669,
+      "loss": 5.0798,
+      "step": 2047
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006817807089859853,
+      "loss": 5.4499,
+      "step": 2048
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006815746084089036,
+      "loss": 5.3822,
+      "step": 2049
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000681368507831822,
+      "loss": 5.1613,
+      "step": 2050
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006811624072547404,
+      "loss": 4.9052,
+      "step": 2051
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006809563066776588,
+      "loss": 5.2296,
+      "step": 2052
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006807502061005772,
+      "loss": 5.1617,
+      "step": 2053
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006805441055234955,
+      "loss": 5.0941,
+      "step": 2054
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006803380049464139,
+      "loss": 5.212,
+      "step": 2055
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006801319043693323,
+      "loss": 4.9523,
+      "step": 2056
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006799258037922507,
+      "loss": 4.8858,
+      "step": 2057
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006797197032151691,
+      "loss": 4.9201,
+      "step": 2058
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006795136026380874,
+      "loss": 5.3762,
+      "step": 2059
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0006793075020610058,
+      "loss": 5.0641,
+      "step": 2060
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006791014014839242,
+      "loss": 5.6986,
+      "step": 2061
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006788953009068426,
+      "loss": 5.5711,
+      "step": 2062
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006786892003297609,
+      "loss": 5.0142,
+      "step": 2063
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006784830997526793,
+      "loss": 4.7655,
+      "step": 2064
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006782769991755977,
+      "loss": 5.2114,
+      "step": 2065
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006780708985985161,
+      "loss": 6.2288,
+      "step": 2066
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006778647980214346,
+      "loss": 4.7475,
+      "step": 2067
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006776586974443528,
+      "loss": 5.1808,
+      "step": 2068
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006774525968672713,
+      "loss": 5.1426,
+      "step": 2069
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006772464962901897,
+      "loss": 5.2208,
+      "step": 2070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006770403957131081,
+      "loss": 5.1144,
+      "step": 2071
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006768342951360264,
+      "loss": 5.6948,
+      "step": 2072
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006766281945589448,
+      "loss": 4.7862,
+      "step": 2073
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006764220939818632,
+      "loss": 4.8242,
+      "step": 2074
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006762159934047816,
+      "loss": 5.1559,
+      "step": 2075
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006760098928277,
+      "loss": 5.3772,
+      "step": 2076
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006758037922506183,
+      "loss": 6.4152,
+      "step": 2077
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0006755976916735367,
+      "loss": 5.2892,
+      "step": 2078
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006753915910964551,
+      "loss": 4.6184,
+      "step": 2079
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006751854905193735,
+      "loss": 6.2252,
+      "step": 2080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006749793899422918,
+      "loss": 5.2546,
+      "step": 2081
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006747732893652102,
+      "loss": 4.834,
+      "step": 2082
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006745671887881286,
+      "loss": 5.3901,
+      "step": 2083
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000674361088211047,
+      "loss": 5.1634,
+      "step": 2084
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006741549876339654,
+      "loss": 5.2525,
+      "step": 2085
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006739488870568837,
+      "loss": 4.7525,
+      "step": 2086
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006737427864798021,
+      "loss": 5.5008,
+      "step": 2087
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006735366859027205,
+      "loss": 5.2583,
+      "step": 2088
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000673330585325639,
+      "loss": 4.944,
+      "step": 2089
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006731244847485574,
+      "loss": 5.454,
+      "step": 2090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006729183841714757,
+      "loss": 5.2949,
+      "step": 2091
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006727122835943941,
+      "loss": 5.2396,
+      "step": 2092
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006725061830173125,
+      "loss": 5.3059,
+      "step": 2093
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006723000824402309,
+      "loss": 5.1416,
+      "step": 2094
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006720939818631492,
+      "loss": 5.4315,
+      "step": 2095
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0006718878812860676,
+      "loss": 5.5322,
+      "step": 2096
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000671681780708986,
+      "loss": 5.6894,
+      "step": 2097
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006714756801319044,
+      "loss": 5.7689,
+      "step": 2098
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006712695795548228,
+      "loss": 5.4537,
+      "step": 2099
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006710634789777411,
+      "loss": 4.832,
+      "step": 2100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006708573784006595,
+      "loss": 5.3137,
+      "step": 2101
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006706512778235779,
+      "loss": 5.1139,
+      "step": 2102
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006704451772464963,
+      "loss": 4.9102,
+      "step": 2103
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006702390766694146,
+      "loss": 5.7341,
+      "step": 2104
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000670032976092333,
+      "loss": 4.9982,
+      "step": 2105
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006698268755152514,
+      "loss": 5.16,
+      "step": 2106
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006696207749381698,
+      "loss": 5.3204,
+      "step": 2107
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006694146743610882,
+      "loss": 5.5068,
+      "step": 2108
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006692085737840065,
+      "loss": 5.1442,
+      "step": 2109
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000669002473206925,
+      "loss": 4.9642,
+      "step": 2110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006687963726298434,
+      "loss": 5.313,
+      "step": 2111
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006685902720527618,
+      "loss": 5.6098,
+      "step": 2112
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006683841714756801,
+      "loss": 5.3963,
+      "step": 2113
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0006681780708985985,
+      "loss": 5.2102,
+      "step": 2114
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006679719703215169,
+      "loss": 5.1484,
+      "step": 2115
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006677658697444353,
+      "loss": 4.7878,
+      "step": 2116
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006675597691673537,
+      "loss": 4.7752,
+      "step": 2117
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000667353668590272,
+      "loss": 5.2354,
+      "step": 2118
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006671475680131904,
+      "loss": 4.955,
+      "step": 2119
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006669414674361088,
+      "loss": 5.226,
+      "step": 2120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006667353668590272,
+      "loss": 5.3196,
+      "step": 2121
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006665292662819456,
+      "loss": 5.1008,
+      "step": 2122
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006663231657048639,
+      "loss": 4.982,
+      "step": 2123
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006661170651277823,
+      "loss": 4.8222,
+      "step": 2124
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006659109645507007,
+      "loss": 5.6265,
+      "step": 2125
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006657048639736191,
+      "loss": 5.006,
+      "step": 2126
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006654987633965374,
+      "loss": 4.7798,
+      "step": 2127
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006652926628194558,
+      "loss": 5.1664,
+      "step": 2128
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006650865622423742,
+      "loss": 4.8503,
+      "step": 2129
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006648804616652927,
+      "loss": 5.2968,
+      "step": 2130
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0006646743610882111,
+      "loss": 4.6957,
+      "step": 2131
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006644682605111294,
+      "loss": 5.128,
+      "step": 2132
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006642621599340478,
+      "loss": 5.6227,
+      "step": 2133
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006640560593569662,
+      "loss": 4.5129,
+      "step": 2134
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006638499587798846,
+      "loss": 5.4589,
+      "step": 2135
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006636438582028029,
+      "loss": 5.3849,
+      "step": 2136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006634377576257213,
+      "loss": 5.036,
+      "step": 2137
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006632316570486398,
+      "loss": 4.8701,
+      "step": 2138
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006630255564715582,
+      "loss": 5.0155,
+      "step": 2139
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006628194558944766,
+      "loss": 5.9138,
+      "step": 2140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006626133553173949,
+      "loss": 5.6769,
+      "step": 2141
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006624072547403133,
+      "loss": 4.957,
+      "step": 2142
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006622011541632317,
+      "loss": 5.3121,
+      "step": 2143
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006619950535861501,
+      "loss": 5.2316,
+      "step": 2144
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006617889530090684,
+      "loss": 5.483,
+      "step": 2145
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006615828524319868,
+      "loss": 5.2811,
+      "step": 2146
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006613767518549052,
+      "loss": 5.1328,
+      "step": 2147
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006611706512778237,
+      "loss": 5.0589,
+      "step": 2148
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0006609645507007421,
+      "loss": 5.1638,
+      "step": 2149
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006607584501236604,
+      "loss": 5.1265,
+      "step": 2150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006605523495465788,
+      "loss": 5.5191,
+      "step": 2151
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006603462489694972,
+      "loss": 5.3291,
+      "step": 2152
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006601401483924156,
+      "loss": 5.1916,
+      "step": 2153
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000659934047815334,
+      "loss": 5.5496,
+      "step": 2154
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006597279472382523,
+      "loss": 5.1241,
+      "step": 2155
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006595218466611707,
+      "loss": 4.938,
+      "step": 2156
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006593157460840891,
+      "loss": 5.9729,
+      "step": 2157
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006591096455070075,
+      "loss": 5.1011,
+      "step": 2158
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006589035449299258,
+      "loss": 5.3379,
+      "step": 2159
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006586974443528442,
+      "loss": 5.3285,
+      "step": 2160
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006584913437757626,
+      "loss": 5.1386,
+      "step": 2161
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000658285243198681,
+      "loss": 5.2553,
+      "step": 2162
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006580791426215994,
+      "loss": 5.0831,
+      "step": 2163
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006578730420445177,
+      "loss": 5.0336,
+      "step": 2164
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006576669414674361,
+      "loss": 5.0668,
+      "step": 2165
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006574608408903545,
+      "loss": 5.1509,
+      "step": 2166
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0006572547403132729,
+      "loss": 5.3455,
+      "step": 2167
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006570486397361912,
+      "loss": 5.0225,
+      "step": 2168
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006568425391591096,
+      "loss": 5.0694,
+      "step": 2169
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006566364385820281,
+      "loss": 5.2057,
+      "step": 2170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006564303380049465,
+      "loss": 4.9828,
+      "step": 2171
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006562242374278649,
+      "loss": 5.6058,
+      "step": 2172
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006560181368507832,
+      "loss": 5.0268,
+      "step": 2173
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006558120362737016,
+      "loss": 4.9259,
+      "step": 2174
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00065560593569662,
+      "loss": 5.0945,
+      "step": 2175
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006553998351195384,
+      "loss": 4.85,
+      "step": 2176
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006551937345424568,
+      "loss": 4.7045,
+      "step": 2177
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006549876339653751,
+      "loss": 5.4636,
+      "step": 2178
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006547815333882935,
+      "loss": 4.5906,
+      "step": 2179
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006545754328112119,
+      "loss": 5.3528,
+      "step": 2180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006543693322341303,
+      "loss": 5.038,
+      "step": 2181
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006541632316570486,
+      "loss": 5.1364,
+      "step": 2182
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000653957131079967,
+      "loss": 6.3646,
+      "step": 2183
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006537510305028854,
+      "loss": 4.5992,
+      "step": 2184
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0006535449299258038,
+      "loss": 5.2013,
+      "step": 2185
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006533388293487222,
+      "loss": 5.1465,
+      "step": 2186
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006531327287716405,
+      "loss": 4.9887,
+      "step": 2187
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006529266281945589,
+      "loss": 5.0772,
+      "step": 2188
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006527205276174773,
+      "loss": 5.5852,
+      "step": 2189
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006525144270403958,
+      "loss": 5.0793,
+      "step": 2190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000652308326463314,
+      "loss": 5.2357,
+      "step": 2191
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006521022258862325,
+      "loss": 5.6092,
+      "step": 2192
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006518961253091509,
+      "loss": 5.2663,
+      "step": 2193
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006516900247320693,
+      "loss": 5.345,
+      "step": 2194
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006514839241549877,
+      "loss": 5.0005,
+      "step": 2195
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000651277823577906,
+      "loss": 5.1608,
+      "step": 2196
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006510717230008244,
+      "loss": 5.2252,
+      "step": 2197
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006508656224237428,
+      "loss": 5.5773,
+      "step": 2198
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006506595218466612,
+      "loss": 5.1215,
+      "step": 2199
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006504534212695795,
+      "loss": 5.6114,
+      "step": 2200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006502473206924979,
+      "loss": 5.4344,
+      "step": 2201
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006500412201154163,
+      "loss": 5.1885,
+      "step": 2202
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0006498351195383347,
+      "loss": 4.9648,
+      "step": 2203
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006496290189612531,
+      "loss": 5.3115,
+      "step": 2204
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006494229183841714,
+      "loss": 5.321,
+      "step": 2205
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006492168178070898,
+      "loss": 5.0968,
+      "step": 2206
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006490107172300082,
+      "loss": 5.3893,
+      "step": 2207
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006488046166529266,
+      "loss": 4.9565,
+      "step": 2208
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000648598516075845,
+      "loss": 4.8185,
+      "step": 2209
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006483924154987633,
+      "loss": 4.9724,
+      "step": 2210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006481863149216818,
+      "loss": 5.0279,
+      "step": 2211
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006479802143446002,
+      "loss": 4.9306,
+      "step": 2212
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006477741137675186,
+      "loss": 6.1808,
+      "step": 2213
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006475680131904369,
+      "loss": 5.2119,
+      "step": 2214
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006473619126133553,
+      "loss": 5.0995,
+      "step": 2215
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006471558120362737,
+      "loss": 5.0831,
+      "step": 2216
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006469497114591921,
+      "loss": 6.3324,
+      "step": 2217
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006467436108821105,
+      "loss": 5.2394,
+      "step": 2218
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006465375103050288,
+      "loss": 5.1413,
+      "step": 2219
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006463314097279472,
+      "loss": 5.1425,
+      "step": 2220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0006461253091508656,
+      "loss": 5.7725,
+      "step": 2221
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000645919208573784,
+      "loss": 4.9215,
+      "step": 2222
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006457131079967023,
+      "loss": 4.7739,
+      "step": 2223
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006455070074196207,
+      "loss": 5.7536,
+      "step": 2224
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006453009068425391,
+      "loss": 5.178,
+      "step": 2225
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006450948062654575,
+      "loss": 5.0174,
+      "step": 2226
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006448887056883759,
+      "loss": 5.1698,
+      "step": 2227
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006446826051112942,
+      "loss": 5.1524,
+      "step": 2228
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006444765045342126,
+      "loss": 5.2629,
+      "step": 2229
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000644270403957131,
+      "loss": 5.8344,
+      "step": 2230
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006440643033800496,
+      "loss": 4.7803,
+      "step": 2231
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006438582028029679,
+      "loss": 5.3221,
+      "step": 2232
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006436521022258863,
+      "loss": 5.2753,
+      "step": 2233
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006434460016488047,
+      "loss": 5.0286,
+      "step": 2234
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006432399010717231,
+      "loss": 5.2789,
+      "step": 2235
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006430338004946415,
+      "loss": 4.9456,
+      "step": 2236
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006428276999175598,
+      "loss": 4.8794,
+      "step": 2237
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0006426215993404782,
+      "loss": 5.483,
+      "step": 2238
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006424154987633966,
+      "loss": 5.0662,
+      "step": 2239
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000642209398186315,
+      "loss": 6.4607,
+      "step": 2240
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006420032976092334,
+      "loss": 5.1086,
+      "step": 2241
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006417971970321517,
+      "loss": 6.1536,
+      "step": 2242
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006415910964550701,
+      "loss": 5.1725,
+      "step": 2243
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006413849958779885,
+      "loss": 5.2488,
+      "step": 2244
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006411788953009069,
+      "loss": 5.34,
+      "step": 2245
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006409727947238252,
+      "loss": 5.0794,
+      "step": 2246
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006407666941467436,
+      "loss": 5.0001,
+      "step": 2247
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000640560593569662,
+      "loss": 5.22,
+      "step": 2248
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006403544929925804,
+      "loss": 5.0609,
+      "step": 2249
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006401483924154989,
+      "loss": 4.8006,
+      "step": 2250
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006399422918384172,
+      "loss": 5.3773,
+      "step": 2251
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006397361912613356,
+      "loss": 5.4225,
+      "step": 2252
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000639530090684254,
+      "loss": 5.1711,
+      "step": 2253
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006393239901071724,
+      "loss": 5.6438,
+      "step": 2254
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006391178895300907,
+      "loss": 5.4339,
+      "step": 2255
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0006389117889530091,
+      "loss": 5.4545,
+      "step": 2256
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006387056883759275,
+      "loss": 4.97,
+      "step": 2257
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006384995877988459,
+      "loss": 5.6912,
+      "step": 2258
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006382934872217643,
+      "loss": 5.3119,
+      "step": 2259
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006380873866446826,
+      "loss": 5.8211,
+      "step": 2260
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000637881286067601,
+      "loss": 5.0899,
+      "step": 2261
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006376751854905194,
+      "loss": 4.9097,
+      "step": 2262
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006374690849134378,
+      "loss": 5.3969,
+      "step": 2263
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006372629843363561,
+      "loss": 5.3508,
+      "step": 2264
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006370568837592745,
+      "loss": 5.2461,
+      "step": 2265
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006368507831821929,
+      "loss": 5.2662,
+      "step": 2266
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006366446826051113,
+      "loss": 5.4167,
+      "step": 2267
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006364385820280297,
+      "loss": 5.0069,
+      "step": 2268
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000636232481450948,
+      "loss": 4.8617,
+      "step": 2269
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006360263808738664,
+      "loss": 4.8077,
+      "step": 2270
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006358202802967849,
+      "loss": 5.7248,
+      "step": 2271
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006356141797197033,
+      "loss": 5.0047,
+      "step": 2272
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0006354080791426217,
+      "loss": 5.0256,
+      "step": 2273
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00063520197856554,
+      "loss": 4.6553,
+      "step": 2274
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006349958779884584,
+      "loss": 5.1583,
+      "step": 2275
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006347897774113768,
+      "loss": 4.9231,
+      "step": 2276
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006345836768342952,
+      "loss": 4.6416,
+      "step": 2277
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006343775762572135,
+      "loss": 4.5915,
+      "step": 2278
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006341714756801319,
+      "loss": 5.2825,
+      "step": 2279
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006339653751030503,
+      "loss": 5.0358,
+      "step": 2280
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006337592745259687,
+      "loss": 5.0435,
+      "step": 2281
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006335531739488871,
+      "loss": 5.1448,
+      "step": 2282
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006333470733718054,
+      "loss": 5.1732,
+      "step": 2283
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006331409727947238,
+      "loss": 5.455,
+      "step": 2284
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006329348722176422,
+      "loss": 5.7795,
+      "step": 2285
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006327287716405606,
+      "loss": 5.3107,
+      "step": 2286
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006325226710634789,
+      "loss": 5.1701,
+      "step": 2287
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006323165704863973,
+      "loss": 5.1203,
+      "step": 2288
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006321104699093157,
+      "loss": 5.0208,
+      "step": 2289
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006319043693322341,
+      "loss": 5.0495,
+      "step": 2290
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006316982687551526,
+      "loss": 5.4818,
+      "step": 2291
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0006314921681780709,
+      "loss": 5.3506,
+      "step": 2292
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006312860676009893,
+      "loss": 5.007,
+      "step": 2293
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006310799670239077,
+      "loss": 5.0757,
+      "step": 2294
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006308738664468261,
+      "loss": 5.1971,
+      "step": 2295
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006306677658697445,
+      "loss": 5.2366,
+      "step": 2296
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006304616652926628,
+      "loss": 4.8623,
+      "step": 2297
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006302555647155812,
+      "loss": 5.2875,
+      "step": 2298
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006300494641384996,
+      "loss": 5.1496,
+      "step": 2299
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000629843363561418,
+      "loss": 5.0305,
+      "step": 2300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006296372629843363,
+      "loss": 4.9972,
+      "step": 2301
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006294311624072547,
+      "loss": 4.9817,
+      "step": 2302
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006292250618301731,
+      "loss": 5.4915,
+      "step": 2303
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006290189612530915,
+      "loss": 4.8869,
+      "step": 2304
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006288128606760099,
+      "loss": 5.023,
+      "step": 2305
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006286067600989282,
+      "loss": 6.0107,
+      "step": 2306
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006284006595218466,
+      "loss": 5.6073,
+      "step": 2307
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000628194558944765,
+      "loss": 5.0245,
+      "step": 2308
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006279884583676834,
+      "loss": 4.9227,
+      "step": 2309
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0006277823577906017,
+      "loss": 4.9639,
+      "step": 2310
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006275762572135201,
+      "loss": 5.5747,
+      "step": 2311
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006273701566364386,
+      "loss": 4.8608,
+      "step": 2312
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000627164056059357,
+      "loss": 5.1174,
+      "step": 2313
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006269579554822754,
+      "loss": 5.3976,
+      "step": 2314
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006267518549051937,
+      "loss": 5.0692,
+      "step": 2315
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006265457543281121,
+      "loss": 5.3769,
+      "step": 2316
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006263396537510305,
+      "loss": 5.0922,
+      "step": 2317
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006261335531739489,
+      "loss": 5.6201,
+      "step": 2318
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006259274525968672,
+      "loss": 5.015,
+      "step": 2319
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006257213520197856,
+      "loss": 5.5318,
+      "step": 2320
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000625515251442704,
+      "loss": 5.1364,
+      "step": 2321
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006253091508656224,
+      "loss": 4.8137,
+      "step": 2322
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006251030502885409,
+      "loss": 5.2088,
+      "step": 2323
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006248969497114591,
+      "loss": 4.982,
+      "step": 2324
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006246908491343776,
+      "loss": 5.059,
+      "step": 2325
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000624484748557296,
+      "loss": 4.8153,
+      "step": 2326
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006242786479802144,
+      "loss": 4.8019,
+      "step": 2327
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0006240725474031328,
+      "loss": 4.7276,
+      "step": 2328
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006238664468260511,
+      "loss": 5.1575,
+      "step": 2329
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006236603462489695,
+      "loss": 5.0034,
+      "step": 2330
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000623454245671888,
+      "loss": 4.447,
+      "step": 2331
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006232481450948064,
+      "loss": 4.9644,
+      "step": 2332
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006230420445177247,
+      "loss": 5.3557,
+      "step": 2333
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006228359439406431,
+      "loss": 6.6065,
+      "step": 2334
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006226298433635615,
+      "loss": 5.3519,
+      "step": 2335
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006224237427864799,
+      "loss": 4.8595,
+      "step": 2336
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006222176422093983,
+      "loss": 5.0925,
+      "step": 2337
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006220115416323166,
+      "loss": 5.1535,
+      "step": 2338
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000621805441055235,
+      "loss": 5.055,
+      "step": 2339
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006215993404781534,
+      "loss": 5.2458,
+      "step": 2340
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006213932399010718,
+      "loss": 6.1997,
+      "step": 2341
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006211871393239901,
+      "loss": 5.2191,
+      "step": 2342
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006209810387469085,
+      "loss": 5.1223,
+      "step": 2343
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006207749381698269,
+      "loss": 5.2316,
+      "step": 2344
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0006205688375927453,
+      "loss": 5.0273,
+      "step": 2345
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006203627370156637,
+      "loss": 5.2203,
+      "step": 2346
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000620156636438582,
+      "loss": 5.1828,
+      "step": 2347
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006199505358615004,
+      "loss": 5.0651,
+      "step": 2348
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006197444352844188,
+      "loss": 4.928,
+      "step": 2349
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006195383347073372,
+      "loss": 5.3052,
+      "step": 2350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006193322341302555,
+      "loss": 4.8101,
+      "step": 2351
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000619126133553174,
+      "loss": 4.7453,
+      "step": 2352
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006189200329760924,
+      "loss": 5.0476,
+      "step": 2353
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006187139323990108,
+      "loss": 4.7978,
+      "step": 2354
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006185078318219292,
+      "loss": 5.0969,
+      "step": 2355
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006183017312448475,
+      "loss": 4.9381,
+      "step": 2356
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006180956306677659,
+      "loss": 5.4551,
+      "step": 2357
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006178895300906843,
+      "loss": 4.7322,
+      "step": 2358
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006176834295136027,
+      "loss": 5.1278,
+      "step": 2359
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006174773289365211,
+      "loss": 5.3264,
+      "step": 2360
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006172712283594394,
+      "loss": 5.5118,
+      "step": 2361
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006170651277823578,
+      "loss": 5.197,
+      "step": 2362
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0006168590272052762,
+      "loss": 5.3968,
+      "step": 2363
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006166529266281946,
+      "loss": 5.6095,
+      "step": 2364
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006164468260511129,
+      "loss": 5.4896,
+      "step": 2365
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006162407254740313,
+      "loss": 5.0424,
+      "step": 2366
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006160346248969497,
+      "loss": 5.4709,
+      "step": 2367
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006158285243198681,
+      "loss": 5.5615,
+      "step": 2368
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006156224237427865,
+      "loss": 5.5211,
+      "step": 2369
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006154163231657048,
+      "loss": 5.2162,
+      "step": 2370
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006152102225886232,
+      "loss": 4.9152,
+      "step": 2371
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006150041220115417,
+      "loss": 5.3815,
+      "step": 2372
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006147980214344601,
+      "loss": 4.8902,
+      "step": 2373
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006145919208573784,
+      "loss": 5.0309,
+      "step": 2374
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006143858202802968,
+      "loss": 5.2859,
+      "step": 2375
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006141797197032152,
+      "loss": 5.4372,
+      "step": 2376
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006139736191261336,
+      "loss": 4.9983,
+      "step": 2377
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000613767518549052,
+      "loss": 4.9865,
+      "step": 2378
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006135614179719703,
+      "loss": 4.9476,
+      "step": 2379
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006133553173948887,
+      "loss": 5.158,
+      "step": 2380
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0006131492168178071,
+      "loss": 4.8281,
+      "step": 2381
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006129431162407255,
+      "loss": 5.048,
+      "step": 2382
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006127370156636439,
+      "loss": 5.0198,
+      "step": 2383
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006125309150865622,
+      "loss": 4.9692,
+      "step": 2384
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006123248145094806,
+      "loss": 5.3318,
+      "step": 2385
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000612118713932399,
+      "loss": 5.1504,
+      "step": 2386
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006119126133553174,
+      "loss": 5.1345,
+      "step": 2387
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006117065127782357,
+      "loss": 4.9226,
+      "step": 2388
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006115004122011541,
+      "loss": 5.0381,
+      "step": 2389
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006112943116240725,
+      "loss": 5.4079,
+      "step": 2390
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000611088211046991,
+      "loss": 5.1037,
+      "step": 2391
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006108821104699094,
+      "loss": 5.3251,
+      "step": 2392
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006106760098928277,
+      "loss": 5.2778,
+      "step": 2393
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006104699093157461,
+      "loss": 5.249,
+      "step": 2394
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006102638087386645,
+      "loss": 4.6148,
+      "step": 2395
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006100577081615829,
+      "loss": 4.9112,
+      "step": 2396
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006098516075845012,
+      "loss": 5.0147,
+      "step": 2397
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0006096455070074196,
+      "loss": 4.9376,
+      "step": 2398
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000609439406430338,
+      "loss": 6.4373,
+      "step": 2399
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006092333058532564,
+      "loss": 5.1855,
+      "step": 2400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006090272052761748,
+      "loss": 5.5658,
+      "step": 2401
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006088211046990931,
+      "loss": 4.9581,
+      "step": 2402
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006086150041220115,
+      "loss": 5.8401,
+      "step": 2403
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006084089035449299,
+      "loss": 5.6608,
+      "step": 2404
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006082028029678483,
+      "loss": 5.2911,
+      "step": 2405
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006079967023907666,
+      "loss": 5.0268,
+      "step": 2406
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000607790601813685,
+      "loss": 4.9876,
+      "step": 2407
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006075845012366034,
+      "loss": 5.1235,
+      "step": 2408
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006073784006595218,
+      "loss": 5.3355,
+      "step": 2409
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006071723000824402,
+      "loss": 5.0474,
+      "step": 2410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006069661995053585,
+      "loss": 4.8792,
+      "step": 2411
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006067600989282769,
+      "loss": 4.8373,
+      "step": 2412
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006065539983511954,
+      "loss": 5.604,
+      "step": 2413
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006063478977741138,
+      "loss": 4.9874,
+      "step": 2414
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006061417971970322,
+      "loss": 4.9995,
+      "step": 2415
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0006059356966199505,
+      "loss": 6.2615,
+      "step": 2416
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000605729596042869,
+      "loss": 5.0352,
+      "step": 2417
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006055234954657874,
+      "loss": 5.1491,
+      "step": 2418
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006053173948887058,
+      "loss": 5.2176,
+      "step": 2419
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006051112943116241,
+      "loss": 5.1635,
+      "step": 2420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006049051937345425,
+      "loss": 5.2971,
+      "step": 2421
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006046990931574609,
+      "loss": 5.571,
+      "step": 2422
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006044929925803793,
+      "loss": 4.9025,
+      "step": 2423
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006042868920032977,
+      "loss": 5.0542,
+      "step": 2424
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.000604080791426216,
+      "loss": 5.2021,
+      "step": 2425
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006038746908491344,
+      "loss": 5.0656,
+      "step": 2426
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006036685902720528,
+      "loss": 5.1404,
+      "step": 2427
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006034624896949712,
+      "loss": 5.0345,
+      "step": 2428
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006032563891178895,
+      "loss": 4.8552,
+      "step": 2429
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006030502885408079,
+      "loss": 4.8507,
+      "step": 2430
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006028441879637263,
+      "loss": 5.4352,
+      "step": 2431
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006026380873866448,
+      "loss": 4.6371,
+      "step": 2432
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006024319868095632,
+      "loss": 6.9429,
+      "step": 2433
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006022258862324815,
+      "loss": 4.9617,
+      "step": 2434
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0006020197856553999,
+      "loss": 5.0981,
+      "step": 2435
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006018136850783183,
+      "loss": 6.1203,
+      "step": 2436
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006016075845012367,
+      "loss": 6.0282,
+      "step": 2437
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000601401483924155,
+      "loss": 5.6862,
+      "step": 2438
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006011953833470734,
+      "loss": 5.6571,
+      "step": 2439
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006009892827699918,
+      "loss": 5.0345,
+      "step": 2440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006007831821929102,
+      "loss": 5.2053,
+      "step": 2441
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006005770816158286,
+      "loss": 5.3615,
+      "step": 2442
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006003709810387469,
+      "loss": 5.0477,
+      "step": 2443
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0006001648804616653,
+      "loss": 5.0014,
+      "step": 2444
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005999587798845837,
+      "loss": 5.1073,
+      "step": 2445
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005997526793075021,
+      "loss": 5.6236,
+      "step": 2446
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005995465787304205,
+      "loss": 5.3447,
+      "step": 2447
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005993404781533388,
+      "loss": 5.378,
+      "step": 2448
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005991343775762572,
+      "loss": 5.345,
+      "step": 2449
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005989282769991756,
+      "loss": 5.0034,
+      "step": 2450
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000598722176422094,
+      "loss": 5.4855,
+      "step": 2451
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0005985160758450123,
+      "loss": 5.9251,
+      "step": 2452
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005983099752679308,
+      "loss": 5.264,
+      "step": 2453
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005981038746908492,
+      "loss": 4.9164,
+      "step": 2454
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005978977741137676,
+      "loss": 5.1914,
+      "step": 2455
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000597691673536686,
+      "loss": 4.9528,
+      "step": 2456
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005974855729596043,
+      "loss": 5.4927,
+      "step": 2457
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005972794723825227,
+      "loss": 4.8779,
+      "step": 2458
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005970733718054411,
+      "loss": 5.2655,
+      "step": 2459
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005968672712283595,
+      "loss": 4.7335,
+      "step": 2460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005966611706512778,
+      "loss": 5.0525,
+      "step": 2461
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005964550700741962,
+      "loss": 5.0777,
+      "step": 2462
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005962489694971146,
+      "loss": 5.2359,
+      "step": 2463
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000596042868920033,
+      "loss": 5.0261,
+      "step": 2464
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005958367683429514,
+      "loss": 5.1837,
+      "step": 2465
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005956306677658697,
+      "loss": 4.8664,
+      "step": 2466
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005954245671887881,
+      "loss": 5.079,
+      "step": 2467
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005952184666117065,
+      "loss": 6.2812,
+      "step": 2468
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005950123660346249,
+      "loss": 5.2695,
+      "step": 2469
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0005948062654575432,
+      "loss": 5.0875,
+      "step": 2470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005946001648804616,
+      "loss": 5.0809,
+      "step": 2471
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00059439406430338,
+      "loss": 4.8734,
+      "step": 2472
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005941879637262985,
+      "loss": 4.8549,
+      "step": 2473
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005939818631492169,
+      "loss": 4.7333,
+      "step": 2474
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005937757625721352,
+      "loss": 5.1476,
+      "step": 2475
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005935696619950536,
+      "loss": 5.5716,
+      "step": 2476
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000593363561417972,
+      "loss": 5.2006,
+      "step": 2477
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005931574608408904,
+      "loss": 5.3679,
+      "step": 2478
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005929513602638088,
+      "loss": 4.7087,
+      "step": 2479
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005927452596867271,
+      "loss": 5.3689,
+      "step": 2480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005925391591096455,
+      "loss": 4.7802,
+      "step": 2481
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005923330585325639,
+      "loss": 4.7788,
+      "step": 2482
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005921269579554823,
+      "loss": 4.9392,
+      "step": 2483
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005919208573784006,
+      "loss": 5.1491,
+      "step": 2484
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000591714756801319,
+      "loss": 5.873,
+      "step": 2485
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005915086562242374,
+      "loss": 5.2447,
+      "step": 2486
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005913025556471558,
+      "loss": 5.1265,
+      "step": 2487
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0005910964550700742,
+      "loss": 5.3334,
+      "step": 2488
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005908903544929925,
+      "loss": 5.2138,
+      "step": 2489
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005906842539159109,
+      "loss": 5.435,
+      "step": 2490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005904781533388293,
+      "loss": 5.2118,
+      "step": 2491
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005902720527617477,
+      "loss": 5.5498,
+      "step": 2492
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000590065952184666,
+      "loss": 5.0135,
+      "step": 2493
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005898598516075845,
+      "loss": 4.953,
+      "step": 2494
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005896537510305029,
+      "loss": 4.9082,
+      "step": 2495
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005894476504534213,
+      "loss": 5.0887,
+      "step": 2496
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005892415498763397,
+      "loss": 5.1033,
+      "step": 2497
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000589035449299258,
+      "loss": 5.2659,
+      "step": 2498
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005888293487221764,
+      "loss": 4.9563,
+      "step": 2499
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005886232481450948,
+      "loss": 5.1524,
+      "step": 2500
+    },
+    {
+      "epoch": 1.4,
+      "eval_loss": 5.525964736938477,
+      "eval_runtime": 1254.8329,
+      "eval_samples_per_second": 2.105,
+      "eval_steps_per_second": 0.264,
+      "eval_wer": 1.3352836176120586,
+      "step": 2500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005884171475680132,
+      "loss": 4.8236,
+      "step": 2501
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005882110469909316,
+      "loss": 5.2301,
+      "step": 2502
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005880049464138499,
+      "loss": 5.0744,
+      "step": 2503
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005877988458367683,
+      "loss": 4.8406,
+      "step": 2504
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005875927452596867,
+      "loss": 4.7324,
+      "step": 2505
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0005873866446826051,
+      "loss": 5.1158,
+      "step": 2506
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005871805441055234,
+      "loss": 4.8253,
+      "step": 2507
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005869744435284418,
+      "loss": 4.9894,
+      "step": 2508
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005867683429513602,
+      "loss": 4.8246,
+      "step": 2509
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005865622423742787,
+      "loss": 5.754,
+      "step": 2510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005863561417971971,
+      "loss": 4.8241,
+      "step": 2511
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005861500412201154,
+      "loss": 4.9667,
+      "step": 2512
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005859439406430339,
+      "loss": 4.7028,
+      "step": 2513
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005857378400659523,
+      "loss": 4.9586,
+      "step": 2514
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005855317394888707,
+      "loss": 5.4356,
+      "step": 2515
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000585325638911789,
+      "loss": 4.9147,
+      "step": 2516
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005851195383347074,
+      "loss": 5.0938,
+      "step": 2517
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005849134377576258,
+      "loss": 5.4864,
+      "step": 2518
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005847073371805442,
+      "loss": 5.2653,
+      "step": 2519
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005845012366034626,
+      "loss": 5.1156,
+      "step": 2520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005842951360263809,
+      "loss": 5.0515,
+      "step": 2521
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005840890354492993,
+      "loss": 4.924,
+      "step": 2522
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005838829348722177,
+      "loss": 5.1704,
+      "step": 2523
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0005836768342951361,
+      "loss": 5.3869,
+      "step": 2524
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005834707337180544,
+      "loss": 4.7542,
+      "step": 2525
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005832646331409728,
+      "loss": 5.2423,
+      "step": 2526
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005830585325638912,
+      "loss": 5.2343,
+      "step": 2527
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005828524319868096,
+      "loss": 4.8419,
+      "step": 2528
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000582646331409728,
+      "loss": 4.8156,
+      "step": 2529
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005824402308326463,
+      "loss": 5.4914,
+      "step": 2530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005822341302555647,
+      "loss": 5.1073,
+      "step": 2531
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005820280296784831,
+      "loss": 5.8907,
+      "step": 2532
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005818219291014016,
+      "loss": 5.3155,
+      "step": 2533
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00058161582852432,
+      "loss": 5.2848,
+      "step": 2534
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005814097279472383,
+      "loss": 4.9392,
+      "step": 2535
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005812036273701567,
+      "loss": 5.4218,
+      "step": 2536
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005809975267930751,
+      "loss": 5.5632,
+      "step": 2537
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005807914262159935,
+      "loss": 5.6222,
+      "step": 2538
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005805853256389118,
+      "loss": 4.9779,
+      "step": 2539
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005803792250618302,
+      "loss": 5.1455,
+      "step": 2540
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0005801731244847486,
+      "loss": 4.9588,
+      "step": 2541
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000579967023907667,
+      "loss": 5.16,
+      "step": 2542
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005797609233305854,
+      "loss": 4.9825,
+      "step": 2543
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005795548227535037,
+      "loss": 4.9513,
+      "step": 2544
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005793487221764221,
+      "loss": 5.2305,
+      "step": 2545
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005791426215993405,
+      "loss": 5.1951,
+      "step": 2546
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005789365210222589,
+      "loss": 5.1869,
+      "step": 2547
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005787304204451772,
+      "loss": 5.0627,
+      "step": 2548
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005785243198680956,
+      "loss": 4.8099,
+      "step": 2549
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000578318219291014,
+      "loss": 4.9451,
+      "step": 2550
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005781121187139324,
+      "loss": 4.9006,
+      "step": 2551
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005779060181368508,
+      "loss": 5.2246,
+      "step": 2552
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005776999175597691,
+      "loss": 5.0998,
+      "step": 2553
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005774938169826876,
+      "loss": 5.5108,
+      "step": 2554
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000577287716405606,
+      "loss": 5.19,
+      "step": 2555
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005770816158285244,
+      "loss": 4.6973,
+      "step": 2556
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005768755152514427,
+      "loss": 5.0565,
+      "step": 2557
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005766694146743611,
+      "loss": 4.8122,
+      "step": 2558
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005764633140972795,
+      "loss": 5.473,
+      "step": 2559
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0005762572135201979,
+      "loss": 5.1547,
+      "step": 2560
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005760511129431163,
+      "loss": 4.8828,
+      "step": 2561
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005758450123660346,
+      "loss": 4.9848,
+      "step": 2562
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000575638911788953,
+      "loss": 4.9934,
+      "step": 2563
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005754328112118714,
+      "loss": 5.0746,
+      "step": 2564
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005752267106347898,
+      "loss": 4.8456,
+      "step": 2565
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005750206100577082,
+      "loss": 4.9453,
+      "step": 2566
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005748145094806265,
+      "loss": 5.0134,
+      "step": 2567
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005746084089035449,
+      "loss": 4.8803,
+      "step": 2568
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005744023083264633,
+      "loss": 5.2943,
+      "step": 2569
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005741962077493817,
+      "loss": 5.0227,
+      "step": 2570
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005739901071723,
+      "loss": 5.0192,
+      "step": 2571
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005737840065952184,
+      "loss": 4.9712,
+      "step": 2572
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005735779060181368,
+      "loss": 5.9621,
+      "step": 2573
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005733718054410553,
+      "loss": 5.332,
+      "step": 2574
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005731657048639737,
+      "loss": 4.5763,
+      "step": 2575
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000572959604286892,
+      "loss": 5.3704,
+      "step": 2576
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0005727535037098104,
+      "loss": 5.5537,
+      "step": 2577
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005725474031327288,
+      "loss": 4.7861,
+      "step": 2578
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005723413025556472,
+      "loss": 4.6511,
+      "step": 2579
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005721352019785655,
+      "loss": 4.9781,
+      "step": 2580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005719291014014839,
+      "loss": 5.5899,
+      "step": 2581
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005717230008244023,
+      "loss": 6.4333,
+      "step": 2582
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005715169002473207,
+      "loss": 4.7318,
+      "step": 2583
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005713107996702391,
+      "loss": 5.2856,
+      "step": 2584
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005711046990931574,
+      "loss": 5.0459,
+      "step": 2585
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005708985985160758,
+      "loss": 5.2426,
+      "step": 2586
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005706924979389942,
+      "loss": 5.0915,
+      "step": 2587
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005704863973619126,
+      "loss": 5.0366,
+      "step": 2588
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005702802967848309,
+      "loss": 5.0057,
+      "step": 2589
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005700741962077493,
+      "loss": 5.0198,
+      "step": 2590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005698680956306677,
+      "loss": 4.8647,
+      "step": 2591
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005696619950535861,
+      "loss": 5.0871,
+      "step": 2592
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005694558944765045,
+      "loss": 5.0787,
+      "step": 2593
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005692497938994228,
+      "loss": 4.999,
+      "step": 2594
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0005690436933223412,
+      "loss": 5.1929,
+      "step": 2595
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005688375927452597,
+      "loss": 4.9962,
+      "step": 2596
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005686314921681781,
+      "loss": 5.0722,
+      "step": 2597
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005684253915910965,
+      "loss": 5.2913,
+      "step": 2598
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005682192910140148,
+      "loss": 4.8036,
+      "step": 2599
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005680131904369332,
+      "loss": 5.1265,
+      "step": 2600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005678070898598516,
+      "loss": 5.024,
+      "step": 2601
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00056760098928277,
+      "loss": 4.9944,
+      "step": 2602
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005673948887056883,
+      "loss": 5.1641,
+      "step": 2603
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005671887881286068,
+      "loss": 5.0299,
+      "step": 2604
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005669826875515252,
+      "loss": 5.4884,
+      "step": 2605
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005667765869744436,
+      "loss": 4.788,
+      "step": 2606
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000566570486397362,
+      "loss": 4.8665,
+      "step": 2607
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005663643858202803,
+      "loss": 4.9571,
+      "step": 2608
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005661582852431987,
+      "loss": 4.7718,
+      "step": 2609
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005659521846661171,
+      "loss": 5.8599,
+      "step": 2610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005657460840890355,
+      "loss": 5.0595,
+      "step": 2611
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005655399835119538,
+      "loss": 5.0144,
+      "step": 2612
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0005653338829348722,
+      "loss": 5.2013,
+      "step": 2613
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005651277823577907,
+      "loss": 4.972,
+      "step": 2614
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005649216817807091,
+      "loss": 5.3148,
+      "step": 2615
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005647155812036275,
+      "loss": 5.0326,
+      "step": 2616
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005645094806265458,
+      "loss": 4.8735,
+      "step": 2617
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005643033800494642,
+      "loss": 4.6936,
+      "step": 2618
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005640972794723826,
+      "loss": 5.0597,
+      "step": 2619
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000563891178895301,
+      "loss": 4.7188,
+      "step": 2620
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005636850783182194,
+      "loss": 4.9935,
+      "step": 2621
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005634789777411377,
+      "loss": 4.9742,
+      "step": 2622
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005632728771640561,
+      "loss": 4.886,
+      "step": 2623
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005630667765869745,
+      "loss": 4.949,
+      "step": 2624
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005628606760098929,
+      "loss": 5.0757,
+      "step": 2625
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005626545754328112,
+      "loss": 4.8529,
+      "step": 2626
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005624484748557296,
+      "loss": 5.0287,
+      "step": 2627
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000562242374278648,
+      "loss": 5.2964,
+      "step": 2628
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005620362737015664,
+      "loss": 4.7067,
+      "step": 2629
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005618301731244848,
+      "loss": 5.1043,
+      "step": 2630
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0005616240725474031,
+      "loss": 5.0189,
+      "step": 2631
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005614179719703215,
+      "loss": 5.0599,
+      "step": 2632
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005612118713932399,
+      "loss": 4.9188,
+      "step": 2633
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005610057708161584,
+      "loss": 4.5415,
+      "step": 2634
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005607996702390767,
+      "loss": 4.8996,
+      "step": 2635
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005605935696619951,
+      "loss": 5.0368,
+      "step": 2636
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005603874690849135,
+      "loss": 5.1274,
+      "step": 2637
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005601813685078319,
+      "loss": 4.9557,
+      "step": 2638
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005599752679307503,
+      "loss": 4.9303,
+      "step": 2639
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005597691673536686,
+      "loss": 5.2072,
+      "step": 2640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000559563066776587,
+      "loss": 5.0132,
+      "step": 2641
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005593569661995054,
+      "loss": 4.9931,
+      "step": 2642
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005591508656224238,
+      "loss": 4.9869,
+      "step": 2643
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005589447650453421,
+      "loss": 4.9798,
+      "step": 2644
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005587386644682605,
+      "loss": 5.1056,
+      "step": 2645
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005585325638911789,
+      "loss": 5.0019,
+      "step": 2646
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005583264633140973,
+      "loss": 5.3082,
+      "step": 2647
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0005581203627370157,
+      "loss": 4.7489,
+      "step": 2648
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000557914262159934,
+      "loss": 5.0893,
+      "step": 2649
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005577081615828524,
+      "loss": 4.8953,
+      "step": 2650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005575020610057708,
+      "loss": 5.2701,
+      "step": 2651
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005572959604286892,
+      "loss": 5.0088,
+      "step": 2652
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005570898598516076,
+      "loss": 4.9419,
+      "step": 2653
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005568837592745259,
+      "loss": 5.2796,
+      "step": 2654
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005566776586974444,
+      "loss": 4.7368,
+      "step": 2655
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005564715581203628,
+      "loss": 5.3595,
+      "step": 2656
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005562654575432812,
+      "loss": 4.977,
+      "step": 2657
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005560593569661995,
+      "loss": 4.8947,
+      "step": 2658
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005558532563891179,
+      "loss": 5.0827,
+      "step": 2659
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005556471558120363,
+      "loss": 4.9061,
+      "step": 2660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005554410552349547,
+      "loss": 5.0193,
+      "step": 2661
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005552349546578731,
+      "loss": 5.1399,
+      "step": 2662
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005550288540807914,
+      "loss": 5.5544,
+      "step": 2663
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005548227535037098,
+      "loss": 4.948,
+      "step": 2664
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005546166529266282,
+      "loss": 5.1481,
+      "step": 2665
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005544105523495466,
+      "loss": 4.9575,
+      "step": 2666
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0005542044517724649,
+      "loss": 5.5299,
+      "step": 2667
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005539983511953833,
+      "loss": 5.04,
+      "step": 2668
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005537922506183017,
+      "loss": 5.2467,
+      "step": 2669
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005535861500412201,
+      "loss": 5.2492,
+      "step": 2670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005533800494641385,
+      "loss": 5.3568,
+      "step": 2671
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005531739488870568,
+      "loss": 5.2213,
+      "step": 2672
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005529678483099752,
+      "loss": 5.0069,
+      "step": 2673
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005527617477328936,
+      "loss": 4.9399,
+      "step": 2674
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000552555647155812,
+      "loss": 5.6874,
+      "step": 2675
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005523495465787303,
+      "loss": 5.5379,
+      "step": 2676
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005521434460016488,
+      "loss": 5.0512,
+      "step": 2677
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005519373454245672,
+      "loss": 5.3238,
+      "step": 2678
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005517312448474856,
+      "loss": 5.1545,
+      "step": 2679
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000551525144270404,
+      "loss": 4.6742,
+      "step": 2680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005513190436933223,
+      "loss": 4.8295,
+      "step": 2681
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005511129431162407,
+      "loss": 4.8877,
+      "step": 2682
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005509068425391591,
+      "loss": 5.1293,
+      "step": 2683
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0005507007419620775,
+      "loss": 4.3582,
+      "step": 2684
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005504946413849959,
+      "loss": 5.4824,
+      "step": 2685
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005502885408079142,
+      "loss": 4.8742,
+      "step": 2686
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005500824402308326,
+      "loss": 4.9243,
+      "step": 2687
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000549876339653751,
+      "loss": 5.0622,
+      "step": 2688
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005496702390766694,
+      "loss": 5.1213,
+      "step": 2689
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005494641384995877,
+      "loss": 4.9794,
+      "step": 2690
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005492580379225061,
+      "loss": 4.8723,
+      "step": 2691
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005490519373454245,
+      "loss": 5.2955,
+      "step": 2692
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005488458367683429,
+      "loss": 5.179,
+      "step": 2693
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005486397361912613,
+      "loss": 4.9995,
+      "step": 2694
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005484336356141796,
+      "loss": 5.2342,
+      "step": 2695
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000548227535037098,
+      "loss": 4.9476,
+      "step": 2696
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005480214344600166,
+      "loss": 4.8588,
+      "step": 2697
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000547815333882935,
+      "loss": 4.7413,
+      "step": 2698
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005476092333058533,
+      "loss": 5.1188,
+      "step": 2699
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005474031327287717,
+      "loss": 5.2071,
+      "step": 2700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005471970321516901,
+      "loss": 4.9804,
+      "step": 2701
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0005469909315746085,
+      "loss": 4.5822,
+      "step": 2702
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005467848309975269,
+      "loss": 4.8061,
+      "step": 2703
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005465787304204452,
+      "loss": 5.3125,
+      "step": 2704
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005463726298433636,
+      "loss": 4.9431,
+      "step": 2705
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000546166529266282,
+      "loss": 5.1095,
+      "step": 2706
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005459604286892004,
+      "loss": 4.8985,
+      "step": 2707
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005457543281121187,
+      "loss": 4.9046,
+      "step": 2708
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005455482275350371,
+      "loss": 5.0342,
+      "step": 2709
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005453421269579555,
+      "loss": 4.8282,
+      "step": 2710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005451360263808739,
+      "loss": 4.7558,
+      "step": 2711
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005449299258037923,
+      "loss": 4.9746,
+      "step": 2712
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005447238252267106,
+      "loss": 4.8267,
+      "step": 2713
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000544517724649629,
+      "loss": 5.3936,
+      "step": 2714
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005443116240725475,
+      "loss": 4.6845,
+      "step": 2715
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005441055234954659,
+      "loss": 5.1043,
+      "step": 2716
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005438994229183843,
+      "loss": 5.2201,
+      "step": 2717
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005436933223413026,
+      "loss": 4.8856,
+      "step": 2718
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000543487221764221,
+      "loss": 4.7968,
+      "step": 2719
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0005432811211871394,
+      "loss": 5.2955,
+      "step": 2720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005430750206100578,
+      "loss": 5.1923,
+      "step": 2721
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005428689200329761,
+      "loss": 5.1323,
+      "step": 2722
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005426628194558945,
+      "loss": 5.167,
+      "step": 2723
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005424567188788129,
+      "loss": 4.9947,
+      "step": 2724
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005422506183017313,
+      "loss": 5.0364,
+      "step": 2725
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005420445177246497,
+      "loss": 4.9826,
+      "step": 2726
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000541838417147568,
+      "loss": 5.2116,
+      "step": 2727
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005416323165704864,
+      "loss": 5.0034,
+      "step": 2728
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005414262159934048,
+      "loss": 5.1489,
+      "step": 2729
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005412201154163232,
+      "loss": 5.6767,
+      "step": 2730
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005410140148392415,
+      "loss": 4.7518,
+      "step": 2731
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005408079142621599,
+      "loss": 4.5024,
+      "step": 2732
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005406018136850783,
+      "loss": 5.8844,
+      "step": 2733
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005403957131079967,
+      "loss": 4.4741,
+      "step": 2734
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005401896125309152,
+      "loss": 5.6628,
+      "step": 2735
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005399835119538334,
+      "loss": 5.3117,
+      "step": 2736
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005397774113767519,
+      "loss": 5.1468,
+      "step": 2737
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0005395713107996703,
+      "loss": 5.1147,
+      "step": 2738
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005393652102225887,
+      "loss": 4.9958,
+      "step": 2739
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005391591096455071,
+      "loss": 4.9375,
+      "step": 2740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005389530090684254,
+      "loss": 5.3149,
+      "step": 2741
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005387469084913438,
+      "loss": 5.0053,
+      "step": 2742
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005385408079142622,
+      "loss": 4.9655,
+      "step": 2743
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005383347073371806,
+      "loss": 5.3314,
+      "step": 2744
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005381286067600989,
+      "loss": 5.2101,
+      "step": 2745
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005379225061830173,
+      "loss": 5.1112,
+      "step": 2746
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005377164056059357,
+      "loss": 5.1045,
+      "step": 2747
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005375103050288541,
+      "loss": 5.1892,
+      "step": 2748
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005373042044517725,
+      "loss": 5.6665,
+      "step": 2749
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005370981038746908,
+      "loss": 6.2012,
+      "step": 2750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005368920032976092,
+      "loss": 4.9213,
+      "step": 2751
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005366859027205276,
+      "loss": 4.8094,
+      "step": 2752
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000536479802143446,
+      "loss": 4.8918,
+      "step": 2753
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005362737015663643,
+      "loss": 5.2004,
+      "step": 2754
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005360676009892827,
+      "loss": 5.1246,
+      "step": 2755
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0005358615004122011,
+      "loss": 4.9691,
+      "step": 2756
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005356553998351196,
+      "loss": 5.2962,
+      "step": 2757
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000535449299258038,
+      "loss": 5.6685,
+      "step": 2758
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005352431986809563,
+      "loss": 5.7795,
+      "step": 2759
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005350370981038747,
+      "loss": 5.1695,
+      "step": 2760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005348309975267931,
+      "loss": 5.0287,
+      "step": 2761
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005346248969497115,
+      "loss": 5.1891,
+      "step": 2762
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005344187963726298,
+      "loss": 4.9988,
+      "step": 2763
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005342126957955482,
+      "loss": 5.0279,
+      "step": 2764
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005340065952184666,
+      "loss": 4.6878,
+      "step": 2765
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000533800494641385,
+      "loss": 4.5447,
+      "step": 2766
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005335943940643034,
+      "loss": 4.802,
+      "step": 2767
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005333882934872217,
+      "loss": 5.1025,
+      "step": 2768
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005331821929101401,
+      "loss": 4.4766,
+      "step": 2769
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005329760923330585,
+      "loss": 4.8029,
+      "step": 2770
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005327699917559769,
+      "loss": 5.018,
+      "step": 2771
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005325638911788953,
+      "loss": 4.7918,
+      "step": 2772
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0005323577906018136,
+      "loss": 4.82,
+      "step": 2773
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000532151690024732,
+      "loss": 5.3638,
+      "step": 2774
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005319455894476504,
+      "loss": 4.7126,
+      "step": 2775
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005317394888705688,
+      "loss": 5.1583,
+      "step": 2776
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005315333882934871,
+      "loss": 5.7386,
+      "step": 2777
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005313272877164056,
+      "loss": 5.0325,
+      "step": 2778
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000531121187139324,
+      "loss": 4.8161,
+      "step": 2779
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005309150865622424,
+      "loss": 4.9083,
+      "step": 2780
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005307089859851608,
+      "loss": 4.4131,
+      "step": 2781
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005305028854080791,
+      "loss": 4.6548,
+      "step": 2782
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005302967848309975,
+      "loss": 4.6686,
+      "step": 2783
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005300906842539159,
+      "loss": 4.3968,
+      "step": 2784
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005298845836768343,
+      "loss": 5.3267,
+      "step": 2785
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005296784830997526,
+      "loss": 5.181,
+      "step": 2786
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000529472382522671,
+      "loss": 5.2621,
+      "step": 2787
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005292662819455894,
+      "loss": 5.1961,
+      "step": 2788
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005290601813685078,
+      "loss": 4.7277,
+      "step": 2789
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005288540807914263,
+      "loss": 4.8467,
+      "step": 2790
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0005286479802143446,
+      "loss": 4.8034,
+      "step": 2791
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000528441879637263,
+      "loss": 4.9587,
+      "step": 2792
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005282357790601814,
+      "loss": 4.9573,
+      "step": 2793
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005280296784830998,
+      "loss": 4.976,
+      "step": 2794
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005278235779060181,
+      "loss": 5.0148,
+      "step": 2795
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005276174773289365,
+      "loss": 4.9686,
+      "step": 2796
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000527411376751855,
+      "loss": 5.0341,
+      "step": 2797
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005272052761747734,
+      "loss": 5.4398,
+      "step": 2798
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005269991755976918,
+      "loss": 5.0283,
+      "step": 2799
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005267930750206101,
+      "loss": 5.2427,
+      "step": 2800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005265869744435285,
+      "loss": 4.8441,
+      "step": 2801
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005263808738664469,
+      "loss": 4.8541,
+      "step": 2802
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005261747732893653,
+      "loss": 4.9279,
+      "step": 2803
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005259686727122837,
+      "loss": 5.5863,
+      "step": 2804
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000525762572135202,
+      "loss": 4.9186,
+      "step": 2805
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005255564715581204,
+      "loss": 5.0983,
+      "step": 2806
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005253503709810388,
+      "loss": 5.4108,
+      "step": 2807
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005251442704039572,
+      "loss": 4.9499,
+      "step": 2808
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0005249381698268755,
+      "loss": 4.7904,
+      "step": 2809
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005247320692497939,
+      "loss": 4.5451,
+      "step": 2810
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005245259686727123,
+      "loss": 4.7912,
+      "step": 2811
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005243198680956307,
+      "loss": 5.0578,
+      "step": 2812
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005241137675185491,
+      "loss": 5.1898,
+      "step": 2813
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005239076669414674,
+      "loss": 5.1514,
+      "step": 2814
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005237015663643858,
+      "loss": 4.8141,
+      "step": 2815
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005234954657873042,
+      "loss": 4.876,
+      "step": 2816
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005232893652102227,
+      "loss": 4.7217,
+      "step": 2817
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000523083264633141,
+      "loss": 4.663,
+      "step": 2818
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005228771640560594,
+      "loss": 4.9491,
+      "step": 2819
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005226710634789778,
+      "loss": 4.811,
+      "step": 2820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005224649629018962,
+      "loss": 4.7992,
+      "step": 2821
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005222588623248146,
+      "loss": 5.0882,
+      "step": 2822
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005220527617477329,
+      "loss": 4.9974,
+      "step": 2823
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005218466611706513,
+      "loss": 5.2016,
+      "step": 2824
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005216405605935697,
+      "loss": 5.3629,
+      "step": 2825
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005214344600164881,
+      "loss": 4.8547,
+      "step": 2826
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0005212283594394065,
+      "loss": 5.121,
+      "step": 2827
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005210222588623248,
+      "loss": 5.3837,
+      "step": 2828
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005208161582852432,
+      "loss": 4.9314,
+      "step": 2829
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005206100577081616,
+      "loss": 5.2105,
+      "step": 2830
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00052040395713108,
+      "loss": 4.4313,
+      "step": 2831
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005201978565539983,
+      "loss": 5.0146,
+      "step": 2832
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005199917559769167,
+      "loss": 4.5366,
+      "step": 2833
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005197856553998351,
+      "loss": 5.012,
+      "step": 2834
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005195795548227535,
+      "loss": 5.0212,
+      "step": 2835
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000519373454245672,
+      "loss": 4.8551,
+      "step": 2836
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005191673536685902,
+      "loss": 4.6085,
+      "step": 2837
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005189612530915087,
+      "loss": 4.9642,
+      "step": 2838
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005187551525144271,
+      "loss": 4.9469,
+      "step": 2839
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005185490519373455,
+      "loss": 4.6395,
+      "step": 2840
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005183429513602638,
+      "loss": 5.6557,
+      "step": 2841
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005181368507831822,
+      "loss": 5.1424,
+      "step": 2842
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005179307502061006,
+      "loss": 5.1396,
+      "step": 2843
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000517724649629019,
+      "loss": 5.6788,
+      "step": 2844
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0005175185490519374,
+      "loss": 4.8759,
+      "step": 2845
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005173124484748557,
+      "loss": 4.9677,
+      "step": 2846
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005171063478977741,
+      "loss": 5.2794,
+      "step": 2847
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005169002473206925,
+      "loss": 5.1317,
+      "step": 2848
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005166941467436109,
+      "loss": 4.8792,
+      "step": 2849
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005164880461665292,
+      "loss": 4.9781,
+      "step": 2850
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005162819455894476,
+      "loss": 5.0135,
+      "step": 2851
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000516075845012366,
+      "loss": 4.9602,
+      "step": 2852
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005158697444352844,
+      "loss": 4.8578,
+      "step": 2853
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005156636438582028,
+      "loss": 4.881,
+      "step": 2854
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005154575432811211,
+      "loss": 4.5714,
+      "step": 2855
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005152514427040395,
+      "loss": 4.85,
+      "step": 2856
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000515045342126958,
+      "loss": 4.8926,
+      "step": 2857
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005148392415498764,
+      "loss": 4.9356,
+      "step": 2858
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005146331409727948,
+      "loss": 5.022,
+      "step": 2859
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005144270403957131,
+      "loss": 4.872,
+      "step": 2860
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005142209398186315,
+      "loss": 5.1583,
+      "step": 2861
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005140148392415499,
+      "loss": 5.0598,
+      "step": 2862
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0005138087386644683,
+      "loss": 5.1173,
+      "step": 2863
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005136026380873866,
+      "loss": 5.0769,
+      "step": 2864
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000513396537510305,
+      "loss": 4.772,
+      "step": 2865
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005131904369332234,
+      "loss": 4.9158,
+      "step": 2866
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005129843363561418,
+      "loss": 4.8843,
+      "step": 2867
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005127782357790602,
+      "loss": 4.7384,
+      "step": 2868
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005125721352019785,
+      "loss": 5.2614,
+      "step": 2869
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005123660346248969,
+      "loss": 4.7651,
+      "step": 2870
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005121599340478153,
+      "loss": 4.8514,
+      "step": 2871
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005119538334707337,
+      "loss": 4.9897,
+      "step": 2872
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000511747732893652,
+      "loss": 5.5515,
+      "step": 2873
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005115416323165704,
+      "loss": 5.1245,
+      "step": 2874
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005113355317394888,
+      "loss": 5.3649,
+      "step": 2875
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005111294311624072,
+      "loss": 4.7959,
+      "step": 2876
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005109233305853256,
+      "loss": 4.7337,
+      "step": 2877
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000510717230008244,
+      "loss": 4.9895,
+      "step": 2878
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005105111294311624,
+      "loss": 4.7825,
+      "step": 2879
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005103050288540808,
+      "loss": 4.5654,
+      "step": 2880
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0005100989282769992,
+      "loss": 4.9235,
+      "step": 2881
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005098928276999175,
+      "loss": 5.5092,
+      "step": 2882
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005096867271228359,
+      "loss": 5.0834,
+      "step": 2883
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005094806265457544,
+      "loss": 4.8762,
+      "step": 2884
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005092745259686728,
+      "loss": 5.3284,
+      "step": 2885
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005090684253915912,
+      "loss": 4.862,
+      "step": 2886
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005088623248145095,
+      "loss": 5.066,
+      "step": 2887
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005086562242374279,
+      "loss": 4.9688,
+      "step": 2888
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005084501236603463,
+      "loss": 4.8806,
+      "step": 2889
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005082440230832647,
+      "loss": 4.9988,
+      "step": 2890
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005080379225061831,
+      "loss": 5.5302,
+      "step": 2891
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005078318219291014,
+      "loss": 5.0152,
+      "step": 2892
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005076257213520198,
+      "loss": 5.0716,
+      "step": 2893
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005074196207749382,
+      "loss": 4.8653,
+      "step": 2894
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005072135201978566,
+      "loss": 4.9987,
+      "step": 2895
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005070074196207749,
+      "loss": 4.8369,
+      "step": 2896
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005068013190436933,
+      "loss": 5.1354,
+      "step": 2897
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005065952184666118,
+      "loss": 4.9034,
+      "step": 2898
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0005063891178895302,
+      "loss": 4.7566,
+      "step": 2899
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005061830173124486,
+      "loss": 4.8962,
+      "step": 2900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005059769167353669,
+      "loss": 4.9501,
+      "step": 2901
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005057708161582853,
+      "loss": 5.0352,
+      "step": 2902
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005055647155812037,
+      "loss": 4.893,
+      "step": 2903
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005053586150041221,
+      "loss": 4.965,
+      "step": 2904
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005051525144270404,
+      "loss": 5.0853,
+      "step": 2905
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005049464138499588,
+      "loss": 5.112,
+      "step": 2906
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005047403132728772,
+      "loss": 5.2239,
+      "step": 2907
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005045342126957956,
+      "loss": 5.2342,
+      "step": 2908
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000504328112118714,
+      "loss": 4.8789,
+      "step": 2909
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005041220115416323,
+      "loss": 5.0946,
+      "step": 2910
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005039159109645507,
+      "loss": 4.9886,
+      "step": 2911
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005037098103874691,
+      "loss": 4.6922,
+      "step": 2912
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005035037098103875,
+      "loss": 5.2768,
+      "step": 2913
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005032976092333058,
+      "loss": 5.1768,
+      "step": 2914
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005030915086562242,
+      "loss": 4.8009,
+      "step": 2915
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0005028854080791426,
+      "loss": 4.5457,
+      "step": 2916
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000502679307502061,
+      "loss": 4.7176,
+      "step": 2917
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005024732069249795,
+      "loss": 5.1096,
+      "step": 2918
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005022671063478978,
+      "loss": 4.9771,
+      "step": 2919
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005020610057708162,
+      "loss": 5.0424,
+      "step": 2920
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005018549051937346,
+      "loss": 4.8301,
+      "step": 2921
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000501648804616653,
+      "loss": 4.6719,
+      "step": 2922
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005014427040395714,
+      "loss": 5.019,
+      "step": 2923
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005012366034624897,
+      "loss": 5.0042,
+      "step": 2924
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005010305028854081,
+      "loss": 4.9556,
+      "step": 2925
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005008244023083265,
+      "loss": 5.2384,
+      "step": 2926
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005006183017312449,
+      "loss": 4.9372,
+      "step": 2927
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005004122011541632,
+      "loss": 4.9669,
+      "step": 2928
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005002061005770816,
+      "loss": 5.2161,
+      "step": 2929
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0005,
+      "loss": 5.3324,
+      "step": 2930
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0004997938994229184,
+      "loss": 4.5944,
+      "step": 2931
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0004995877988458368,
+      "loss": 4.9624,
+      "step": 2932
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0004993816982687552,
+      "loss": 6.7475,
+      "step": 2933
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0004991755976916735,
+      "loss": 4.2017,
+      "step": 2934
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004989694971145919,
+      "loss": 4.9054,
+      "step": 2935
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004987633965375103,
+      "loss": 5.648,
+      "step": 2936
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004985572959604287,
+      "loss": 5.1702,
+      "step": 2937
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004983511953833472,
+      "loss": 4.7033,
+      "step": 2938
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004981450948062655,
+      "loss": 4.8833,
+      "step": 2939
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004979389942291839,
+      "loss": 4.8524,
+      "step": 2940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004977328936521023,
+      "loss": 4.9436,
+      "step": 2941
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004975267930750207,
+      "loss": 4.8493,
+      "step": 2942
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000497320692497939,
+      "loss": 5.1094,
+      "step": 2943
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004971145919208574,
+      "loss": 5.0886,
+      "step": 2944
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004969084913437758,
+      "loss": 5.183,
+      "step": 2945
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004967023907666942,
+      "loss": 4.8224,
+      "step": 2946
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004964962901896126,
+      "loss": 4.9344,
+      "step": 2947
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004962901896125309,
+      "loss": 4.8025,
+      "step": 2948
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004960840890354493,
+      "loss": 5.4065,
+      "step": 2949
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004958779884583677,
+      "loss": 4.6567,
+      "step": 2950
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004956718878812861,
+      "loss": 4.886,
+      "step": 2951
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0004954657873042044,
+      "loss": 4.7828,
+      "step": 2952
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004952596867271228,
+      "loss": 4.6763,
+      "step": 2953
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004950535861500412,
+      "loss": 5.0376,
+      "step": 2954
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004948474855729596,
+      "loss": 4.7115,
+      "step": 2955
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000494641384995878,
+      "loss": 5.5865,
+      "step": 2956
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004944352844187963,
+      "loss": 4.5278,
+      "step": 2957
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004942291838417147,
+      "loss": 5.0636,
+      "step": 2958
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004940230832646332,
+      "loss": 5.0431,
+      "step": 2959
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004938169826875516,
+      "loss": 4.8465,
+      "step": 2960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004936108821104699,
+      "loss": 4.6694,
+      "step": 2961
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004934047815333883,
+      "loss": 4.6394,
+      "step": 2962
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004931986809563067,
+      "loss": 5.2081,
+      "step": 2963
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004929925803792251,
+      "loss": 5.1752,
+      "step": 2964
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004927864798021435,
+      "loss": 4.8374,
+      "step": 2965
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004925803792250618,
+      "loss": 4.7156,
+      "step": 2966
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004923742786479802,
+      "loss": 4.792,
+      "step": 2967
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004921681780708986,
+      "loss": 4.7624,
+      "step": 2968
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000491962077493817,
+      "loss": 4.9026,
+      "step": 2969
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0004917559769167354,
+      "loss": 5.442,
+      "step": 2970
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004915498763396537,
+      "loss": 5.2256,
+      "step": 2971
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004913437757625721,
+      "loss": 4.9746,
+      "step": 2972
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004911376751854905,
+      "loss": 4.8717,
+      "step": 2973
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004909315746084089,
+      "loss": 5.1687,
+      "step": 2974
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004907254740313272,
+      "loss": 4.7791,
+      "step": 2975
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004905193734542456,
+      "loss": 5.0138,
+      "step": 2976
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.000490313272877164,
+      "loss": 5.1594,
+      "step": 2977
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004901071723000824,
+      "loss": 5.3393,
+      "step": 2978
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004899010717230009,
+      "loss": 5.0187,
+      "step": 2979
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004896949711459193,
+      "loss": 4.886,
+      "step": 2980
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004894888705688377,
+      "loss": 4.9279,
+      "step": 2981
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.000489282769991756,
+      "loss": 5.0332,
+      "step": 2982
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004890766694146744,
+      "loss": 5.1307,
+      "step": 2983
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004888705688375928,
+      "loss": 5.2083,
+      "step": 2984
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004886644682605112,
+      "loss": 4.9953,
+      "step": 2985
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004884583676834296,
+      "loss": 5.0148,
+      "step": 2986
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004882522671063479,
+      "loss": 4.8524,
+      "step": 2987
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0004880461665292663,
+      "loss": 4.7368,
+      "step": 2988
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00048784006595218465,
+      "loss": 5.2136,
+      "step": 2989
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00048763396537510306,
+      "loss": 5.1513,
+      "step": 2990
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0004874278647980214,
+      "loss": 5.3179,
+      "step": 2991
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0004872217642209398,
+      "loss": 4.8305,
+      "step": 2992
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00048701566364385817,
+      "loss": 5.357,
+      "step": 2993
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0004868095630667766,
+      "loss": 4.8435,
+      "step": 2994
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000486603462489695,
+      "loss": 5.0059,
+      "step": 2995
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00048639736191261333,
+      "loss": 4.895,
+      "step": 2996
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00048619126133553174,
+      "loss": 5.0743,
+      "step": 2997
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0004859851607584501,
+      "loss": 4.911,
+      "step": 2998
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0004857790601813685,
+      "loss": 4.7798,
+      "step": 2999
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00048557295960428685,
+      "loss": 4.8291,
+      "step": 3000
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 5.403041839599609,
+      "eval_runtime": 1248.2543,
+      "eval_samples_per_second": 2.117,
+      "eval_steps_per_second": 0.265,
+      "eval_wer": 1.9693573978579928,
+      "step": 3000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0004853668590272053,
+      "loss": 4.8723,
+      "step": 3001
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0004851607584501237,
+      "loss": 5.0038,
+      "step": 3002
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00048495465787304207,
+      "loss": 4.7439,
+      "step": 3003
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0004847485572959605,
+      "loss": 4.868,
+      "step": 3004
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00048454245671887883,
+      "loss": 4.7873,
+      "step": 3005
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00048433635614179724,
+      "loss": 5.1154,
+      "step": 3006
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0004841302555647156,
+      "loss": 5.3444,
+      "step": 3007
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.000483924154987634,
+      "loss": 4.7498,
+      "step": 3008
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00048371805441055235,
+      "loss": 4.847,
+      "step": 3009
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00048351195383347076,
+      "loss": 4.8491,
+      "step": 3010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00048330585325638916,
+      "loss": 5.0009,
+      "step": 3011
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0004830997526793075,
+      "loss": 4.617,
+      "step": 3012
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0004828936521022259,
+      "loss": 5.6907,
+      "step": 3013
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0004826875515251443,
+      "loss": 5.0526,
+      "step": 3014
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0004824814509480627,
+      "loss": 4.6219,
+      "step": 3015
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00048227535037098103,
+      "loss": 5.2314,
+      "step": 3016
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00048206924979389944,
+      "loss": 5.8462,
+      "step": 3017
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00048186314921681785,
+      "loss": 4.8529,
+      "step": 3018
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0004816570486397362,
+      "loss": 5.0535,
+      "step": 3019
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0004814509480626546,
+      "loss": 4.8046,
+      "step": 3020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00048124484748557296,
+      "loss": 4.9137,
+      "step": 3021
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00048103874690849137,
+      "loss": 4.697,
+      "step": 3022
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0004808326463314097,
+      "loss": 5.289,
+      "step": 3023
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0004806265457543281,
+      "loss": 4.7171,
+      "step": 3024
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0004804204451772465,
+      "loss": 5.0162,
+      "step": 3025
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0004802143446001649,
+      "loss": 5.0076,
+      "step": 3026
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0004800082440230833,
+      "loss": 4.7561,
+      "step": 3027
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00047980214344600164,
+      "loss": 6.1056,
+      "step": 3028
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00047959604286892005,
+      "loss": 4.8736,
+      "step": 3029
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0004793899422918384,
+      "loss": 6.1025,
+      "step": 3030
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0004791838417147568,
+      "loss": 4.9083,
+      "step": 3031
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00047897774113767516,
+      "loss": 4.9908,
+      "step": 3032
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00047877164056059357,
+      "loss": 5.2447,
+      "step": 3033
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000478565539983512,
+      "loss": 4.6261,
+      "step": 3034
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00047835943940643033,
+      "loss": 4.9987,
+      "step": 3035
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00047815333882934873,
+      "loss": 4.8148,
+      "step": 3036
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0004779472382522671,
+      "loss": 4.9428,
+      "step": 3037
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0004777411376751855,
+      "loss": 4.9927,
+      "step": 3038
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00047753503709810385,
+      "loss": 4.9761,
+      "step": 3039
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00047732893652102225,
+      "loss": 4.9002,
+      "step": 3040
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00047712283594394066,
+      "loss": 4.9317,
+      "step": 3041
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.000476916735366859,
+      "loss": 4.7772,
+      "step": 3042
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004767106347897774,
+      "loss": 4.8361,
+      "step": 3043
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00047650453421269577,
+      "loss": 4.8621,
+      "step": 3044
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004762984336356142,
+      "loss": 4.8283,
+      "step": 3045
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00047609233305853253,
+      "loss": 4.8489,
+      "step": 3046
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.000475886232481451,
+      "loss": 4.7972,
+      "step": 3047
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00047568013190436934,
+      "loss": 5.1444,
+      "step": 3048
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00047547403132728775,
+      "loss": 4.8729,
+      "step": 3049
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00047526793075020616,
+      "loss": 4.5881,
+      "step": 3050
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004750618301731245,
+      "loss": 4.7556,
+      "step": 3051
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004748557295960429,
+      "loss": 4.996,
+      "step": 3052
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00047464962901896127,
+      "loss": 4.9801,
+      "step": 3053
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004744435284418797,
+      "loss": 4.9068,
+      "step": 3054
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00047423742786479803,
+      "loss": 5.1295,
+      "step": 3055
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00047403132728771643,
+      "loss": 4.7982,
+      "step": 3056
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00047382522671063484,
+      "loss": 5.0604,
+      "step": 3057
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004736191261335532,
+      "loss": 5.0209,
+      "step": 3058
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0004734130255564716,
+      "loss": 4.6368,
+      "step": 3059
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00047320692497938995,
+      "loss": 4.7851,
+      "step": 3060
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00047300082440230836,
+      "loss": 4.608,
+      "step": 3061
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0004727947238252267,
+      "loss": 4.9432,
+      "step": 3062
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0004725886232481451,
+      "loss": 5.014,
+      "step": 3063
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00047238252267106347,
+      "loss": 4.9938,
+      "step": 3064
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0004721764220939819,
+      "loss": 4.4423,
+      "step": 3065
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0004719703215169003,
+      "loss": 5.0557,
+      "step": 3066
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00047176422093981864,
+      "loss": 4.7186,
+      "step": 3067
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00047155812036273704,
+      "loss": 5.2847,
+      "step": 3068
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0004713520197856554,
+      "loss": 5.3346,
+      "step": 3069
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0004711459192085738,
+      "loss": 4.9483,
+      "step": 3070
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00047093981863149216,
+      "loss": 5.0191,
+      "step": 3071
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00047073371805441056,
+      "loss": 4.8168,
+      "step": 3072
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00047052761747732897,
+      "loss": 5.0092,
+      "step": 3073
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0004703215169002473,
+      "loss": 4.7912,
+      "step": 3074
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00047011541632316573,
+      "loss": 5.2722,
+      "step": 3075
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0004699093157460841,
+      "loss": 6.0117,
+      "step": 3076
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0004697032151690025,
+      "loss": 4.8997,
+      "step": 3077
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00046949711459192084,
+      "loss": 4.8956,
+      "step": 3078
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00046929101401483925,
+      "loss": 5.1898,
+      "step": 3079
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0004690849134377576,
+      "loss": 5.2208,
+      "step": 3080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000468878812860676,
+      "loss": 4.6402,
+      "step": 3081
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0004686727122835944,
+      "loss": 5.0158,
+      "step": 3082
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00046846661170651276,
+      "loss": 4.7521,
+      "step": 3083
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00046826051112943117,
+      "loss": 6.5034,
+      "step": 3084
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0004680544105523495,
+      "loss": 5.2562,
+      "step": 3085
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00046784830997526793,
+      "loss": 4.9702,
+      "step": 3086
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0004676422093981863,
+      "loss": 5.0026,
+      "step": 3087
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0004674361088211047,
+      "loss": 4.9458,
+      "step": 3088
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0004672300082440231,
+      "loss": 5.027,
+      "step": 3089
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00046702390766694145,
+      "loss": 4.6843,
+      "step": 3090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00046681780708985986,
+      "loss": 5.0988,
+      "step": 3091
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0004666117065127782,
+      "loss": 5.2476,
+      "step": 3092
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0004664056059356966,
+      "loss": 4.7022,
+      "step": 3093
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000466199505358615,
+      "loss": 5.3089,
+      "step": 3094
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00046599340478153343,
+      "loss": 5.2685,
+      "step": 3095
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0004657873042044518,
+      "loss": 5.0519,
+      "step": 3096
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0004655812036273702,
+      "loss": 5.1098,
+      "step": 3097
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0004653751030502886,
+      "loss": 4.6919,
+      "step": 3098
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00046516900247320695,
+      "loss": 4.6333,
+      "step": 3099
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00046496290189612535,
+      "loss": 4.9391,
+      "step": 3100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0004647568013190437,
+      "loss": 5.0169,
+      "step": 3101
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0004645507007419621,
+      "loss": 4.8623,
+      "step": 3102
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00046434460016488046,
+      "loss": 5.1837,
+      "step": 3103
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00046413849958779887,
+      "loss": 4.6672,
+      "step": 3104
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0004639323990107173,
+      "loss": 5.3362,
+      "step": 3105
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00046372629843363563,
+      "loss": 5.1017,
+      "step": 3106
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00046352019785655404,
+      "loss": 4.9099,
+      "step": 3107
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0004633140972794724,
+      "loss": 4.776,
+      "step": 3108
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0004631079967023908,
+      "loss": 4.9726,
+      "step": 3109
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00046290189612530915,
+      "loss": 5.1797,
+      "step": 3110
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00046269579554822756,
+      "loss": 4.9184,
+      "step": 3111
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0004624896949711459,
+      "loss": 5.4449,
+      "step": 3112
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0004622835943940643,
+      "loss": 4.6935,
+      "step": 3113
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004620774938169827,
+      "loss": 4.8471,
+      "step": 3114
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004618713932399011,
+      "loss": 4.946,
+      "step": 3115
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004616652926628195,
+      "loss": 4.7301,
+      "step": 3116
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00046145919208573783,
+      "loss": 4.6532,
+      "step": 3117
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00046125309150865624,
+      "loss": 4.8216,
+      "step": 3118
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004610469909315746,
+      "loss": 4.7028,
+      "step": 3119
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000460840890354493,
+      "loss": 4.7962,
+      "step": 3120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004606347897774114,
+      "loss": 5.1465,
+      "step": 3121
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00046042868920032976,
+      "loss": 4.839,
+      "step": 3122
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00046022258862324817,
+      "loss": 4.8676,
+      "step": 3123
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004600164880461665,
+      "loss": 5.068,
+      "step": 3124
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004598103874690849,
+      "loss": 4.7639,
+      "step": 3125
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004596042868920033,
+      "loss": 4.7681,
+      "step": 3126
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0004593981863149217,
+      "loss": 4.843,
+      "step": 3127
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00045919208573784004,
+      "loss": 4.7507,
+      "step": 3128
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00045898598516075844,
+      "loss": 4.907,
+      "step": 3129
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00045877988458367685,
+      "loss": 4.6486,
+      "step": 3130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004585737840065952,
+      "loss": 4.9815,
+      "step": 3131
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004583676834295136,
+      "loss": 5.6813,
+      "step": 3132
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00045816158285243196,
+      "loss": 4.7036,
+      "step": 3133
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00045795548227535037,
+      "loss": 4.587,
+      "step": 3134
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004577493816982687,
+      "loss": 5.3507,
+      "step": 3135
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004575432811211871,
+      "loss": 5.3118,
+      "step": 3136
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00045733718054410553,
+      "loss": 5.1988,
+      "step": 3137
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004571310799670239,
+      "loss": 4.9579,
+      "step": 3138
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004569249793899423,
+      "loss": 4.9175,
+      "step": 3139
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00045671887881286065,
+      "loss": 5.0173,
+      "step": 3140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004565127782357791,
+      "loss": 5.0964,
+      "step": 3141
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00045630667765869746,
+      "loss": 5.7245,
+      "step": 3142
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00045610057708161587,
+      "loss": 5.1134,
+      "step": 3143
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004558944765045342,
+      "loss": 5.039,
+      "step": 3144
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004556883759274526,
+      "loss": 4.6589,
+      "step": 3145
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00045548227535037103,
+      "loss": 4.5695,
+      "step": 3146
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004552761747732894,
+      "loss": 4.8054,
+      "step": 3147
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0004550700741962078,
+      "loss": 4.9116,
+      "step": 3148
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00045486397361912614,
+      "loss": 4.8478,
+      "step": 3149
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00045465787304204455,
+      "loss": 4.9148,
+      "step": 3150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004544517724649629,
+      "loss": 4.8512,
+      "step": 3151
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004542456718878813,
+      "loss": 5.3454,
+      "step": 3152
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004540395713107997,
+      "loss": 4.9626,
+      "step": 3153
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00045383347073371807,
+      "loss": 5.2522,
+      "step": 3154
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004536273701566365,
+      "loss": 4.8704,
+      "step": 3155
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004534212695795548,
+      "loss": 4.5666,
+      "step": 3156
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00045321516900247323,
+      "loss": 4.994,
+      "step": 3157
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004530090684253916,
+      "loss": 4.8075,
+      "step": 3158
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00045280296784831,
+      "loss": 4.9115,
+      "step": 3159
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004525968672712284,
+      "loss": 5.0906,
+      "step": 3160
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00045239076669414675,
+      "loss": 4.8376,
+      "step": 3161
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00045218466611706516,
+      "loss": 4.6316,
+      "step": 3162
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004519785655399835,
+      "loss": 4.893,
+      "step": 3163
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004517724649629019,
+      "loss": 5.2499,
+      "step": 3164
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00045156636438582027,
+      "loss": 5.204,
+      "step": 3165
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0004513602638087387,
+      "loss": 4.704,
+      "step": 3166
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00045115416323165703,
+      "loss": 5.1799,
+      "step": 3167
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00045094806265457544,
+      "loss": 4.7406,
+      "step": 3168
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00045074196207749384,
+      "loss": 4.7193,
+      "step": 3169
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0004505358615004122,
+      "loss": 4.7475,
+      "step": 3170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0004503297609233306,
+      "loss": 5.1442,
+      "step": 3171
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00045012366034624895,
+      "loss": 4.68,
+      "step": 3172
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00044991755976916736,
+      "loss": 5.0144,
+      "step": 3173
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0004497114591920857,
+      "loss": 4.8541,
+      "step": 3174
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0004495053586150041,
+      "loss": 4.9161,
+      "step": 3175
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0004492992580379225,
+      "loss": 5.0643,
+      "step": 3176
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0004490931574608409,
+      "loss": 5.0593,
+      "step": 3177
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0004488870568837593,
+      "loss": 4.9247,
+      "step": 3178
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00044868095630667764,
+      "loss": 4.8902,
+      "step": 3179
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00044847485572959605,
+      "loss": 5.478,
+      "step": 3180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0004482687551525144,
+      "loss": 5.0754,
+      "step": 3181
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0004480626545754328,
+      "loss": 4.2958,
+      "step": 3182
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00044785655399835116,
+      "loss": 4.8522,
+      "step": 3183
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00044765045342126956,
+      "loss": 4.7896,
+      "step": 3184
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00044744435284418797,
+      "loss": 5.0315,
+      "step": 3185
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0004472382522671063,
+      "loss": 5.0477,
+      "step": 3186
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0004470321516900248,
+      "loss": 4.9975,
+      "step": 3187
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00044682605111294314,
+      "loss": 5.1011,
+      "step": 3188
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00044661995053586154,
+      "loss": 5.2104,
+      "step": 3189
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0004464138499587799,
+      "loss": 5.158,
+      "step": 3190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0004462077493816983,
+      "loss": 4.9534,
+      "step": 3191
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0004460016488046167,
+      "loss": 5.3918,
+      "step": 3192
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00044579554822753506,
+      "loss": 4.9258,
+      "step": 3193
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00044558944765045347,
+      "loss": 4.9855,
+      "step": 3194
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0004453833470733718,
+      "loss": 5.1178,
+      "step": 3195
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00044517724649629023,
+      "loss": 5.2064,
+      "step": 3196
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0004449711459192086,
+      "loss": 4.81,
+      "step": 3197
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000444765045342127,
+      "loss": 5.2041,
+      "step": 3198
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00044455894476504534,
+      "loss": 5.474,
+      "step": 3199
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00044435284418796375,
+      "loss": 5.1858,
+      "step": 3200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00044414674361088215,
+      "loss": 4.7886,
+      "step": 3201
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0004439406430338005,
+      "loss": 4.7581,
+      "step": 3202
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004437345424567189,
+      "loss": 5.0024,
+      "step": 3203
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00044352844187963726,
+      "loss": 5.0007,
+      "step": 3204
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00044332234130255567,
+      "loss": 4.9903,
+      "step": 3205
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000443116240725474,
+      "loss": 5.0942,
+      "step": 3206
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00044291014014839243,
+      "loss": 5.3637,
+      "step": 3207
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00044270403957131084,
+      "loss": 4.9769,
+      "step": 3208
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004424979389942292,
+      "loss": 4.8651,
+      "step": 3209
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004422918384171476,
+      "loss": 5.3516,
+      "step": 3210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00044208573784006595,
+      "loss": 4.9724,
+      "step": 3211
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00044187963726298435,
+      "loss": 4.74,
+      "step": 3212
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004416735366859027,
+      "loss": 4.7992,
+      "step": 3213
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004414674361088211,
+      "loss": 5.1595,
+      "step": 3214
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00044126133553173947,
+      "loss": 5.2401,
+      "step": 3215
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004410552349546579,
+      "loss": 4.8121,
+      "step": 3216
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004408491343775763,
+      "loss": 5.0994,
+      "step": 3217
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00044064303380049463,
+      "loss": 4.7422,
+      "step": 3218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00044043693322341304,
+      "loss": 4.8098,
+      "step": 3219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0004402308326463314,
+      "loss": 4.9356,
+      "step": 3220
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004400247320692498,
+      "loss": 5.0085,
+      "step": 3221
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00043981863149216815,
+      "loss": 4.9139,
+      "step": 3222
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00043961253091508656,
+      "loss": 4.6199,
+      "step": 3223
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00043940643033800496,
+      "loss": 4.8562,
+      "step": 3224
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004392003297609233,
+      "loss": 4.7816,
+      "step": 3225
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004389942291838417,
+      "loss": 5.2963,
+      "step": 3226
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004387881286067601,
+      "loss": 5.9273,
+      "step": 3227
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004385820280296785,
+      "loss": 4.773,
+      "step": 3228
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00043837592745259683,
+      "loss": 4.7837,
+      "step": 3229
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00043816982687551524,
+      "loss": 4.6543,
+      "step": 3230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004379637262984336,
+      "loss": 4.4912,
+      "step": 3231
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000437757625721352,
+      "loss": 4.9077,
+      "step": 3232
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00043755152514427046,
+      "loss": 5.3983,
+      "step": 3233
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004373454245671888,
+      "loss": 4.6457,
+      "step": 3234
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004371393239901072,
+      "loss": 5.6278,
+      "step": 3235
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0004369332234130256,
+      "loss": 4.8786,
+      "step": 3236
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.000436727122835944,
+      "loss": 4.9156,
+      "step": 3237
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00043652102225886233,
+      "loss": 4.8059,
+      "step": 3238
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00043631492168178074,
+      "loss": 4.8107,
+      "step": 3239
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00043610882110469915,
+      "loss": 4.747,
+      "step": 3240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0004359027205276175,
+      "loss": 4.6629,
+      "step": 3241
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0004356966199505359,
+      "loss": 5.2124,
+      "step": 3242
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00043549051937345426,
+      "loss": 5.1492,
+      "step": 3243
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00043528441879637266,
+      "loss": 5.121,
+      "step": 3244
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000435078318219291,
+      "loss": 4.9663,
+      "step": 3245
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0004348722176422094,
+      "loss": 4.919,
+      "step": 3246
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0004346661170651278,
+      "loss": 5.0973,
+      "step": 3247
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0004344600164880462,
+      "loss": 4.7288,
+      "step": 3248
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0004342539159109646,
+      "loss": 4.5941,
+      "step": 3249
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00043404781533388294,
+      "loss": 4.7467,
+      "step": 3250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00043384171475680135,
+      "loss": 4.9226,
+      "step": 3251
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0004336356141797197,
+      "loss": 5.037,
+      "step": 3252
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0004334295136026381,
+      "loss": 5.1349,
+      "step": 3253
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00043322341302555646,
+      "loss": 4.8432,
+      "step": 3254
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00043301731244847487,
+      "loss": 5.0518,
+      "step": 3255
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0004328112118713933,
+      "loss": 4.7045,
+      "step": 3256
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0004326051112943116,
+      "loss": 4.7984,
+      "step": 3257
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00043239901071723003,
+      "loss": 4.9759,
+      "step": 3258
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0004321929101401484,
+      "loss": 4.8902,
+      "step": 3259
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0004319868095630668,
+      "loss": 5.2015,
+      "step": 3260
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00043178070898598514,
+      "loss": 5.1806,
+      "step": 3261
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00043157460840890355,
+      "loss": 5.5116,
+      "step": 3262
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00043136850783182196,
+      "loss": 4.969,
+      "step": 3263
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0004311624072547403,
+      "loss": 4.8411,
+      "step": 3264
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0004309563066776587,
+      "loss": 4.9391,
+      "step": 3265
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00043075020610057707,
+      "loss": 4.7442,
+      "step": 3266
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0004305441055234955,
+      "loss": 4.6708,
+      "step": 3267
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00043033800494641383,
+      "loss": 4.7236,
+      "step": 3268
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00043013190436933224,
+      "loss": 4.9641,
+      "step": 3269
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0004299258037922506,
+      "loss": 5.2206,
+      "step": 3270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000429719703215169,
+      "loss": 4.9477,
+      "step": 3271
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0004295136026380874,
+      "loss": 5.003,
+      "step": 3272
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00042930750206100575,
+      "loss": 4.8477,
+      "step": 3273
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00042910140148392416,
+      "loss": 5.0844,
+      "step": 3274
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004288953009068425,
+      "loss": 6.0636,
+      "step": 3275
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004286892003297609,
+      "loss": 4.9608,
+      "step": 3276
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00042848309975267927,
+      "loss": 4.5444,
+      "step": 3277
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004282769991755977,
+      "loss": 4.9657,
+      "step": 3278
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004280708985985161,
+      "loss": 4.978,
+      "step": 3279
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004278647980214345,
+      "loss": 4.4641,
+      "step": 3280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004276586974443529,
+      "loss": 5.0015,
+      "step": 3281
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00042745259686727125,
+      "loss": 4.3605,
+      "step": 3282
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00042724649629018966,
+      "loss": 5.2178,
+      "step": 3283
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000427040395713108,
+      "loss": 4.2489,
+      "step": 3284
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004268342951360264,
+      "loss": 4.6721,
+      "step": 3285
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00042662819455894477,
+      "loss": 5.2977,
+      "step": 3286
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004264220939818632,
+      "loss": 4.8224,
+      "step": 3287
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004262159934047816,
+      "loss": 4.5592,
+      "step": 3288
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00042600989282769994,
+      "loss": 5.3955,
+      "step": 3289
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00042580379225061834,
+      "loss": 5.0367,
+      "step": 3290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0004255976916735367,
+      "loss": 5.2914,
+      "step": 3291
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004253915910964551,
+      "loss": 4.8524,
+      "step": 3292
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00042518549051937345,
+      "loss": 4.914,
+      "step": 3293
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00042497938994229186,
+      "loss": 4.7175,
+      "step": 3294
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00042477328936521027,
+      "loss": 4.8798,
+      "step": 3295
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004245671887881286,
+      "loss": 4.821,
+      "step": 3296
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000424361088211047,
+      "loss": 4.9643,
+      "step": 3297
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004241549876339654,
+      "loss": 4.8609,
+      "step": 3298
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004239488870568838,
+      "loss": 5.1429,
+      "step": 3299
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00042374278647980214,
+      "loss": 4.7018,
+      "step": 3300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00042353668590272054,
+      "loss": 4.8967,
+      "step": 3301
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004233305853256389,
+      "loss": 4.6832,
+      "step": 3302
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004231244847485573,
+      "loss": 4.9114,
+      "step": 3303
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004229183841714757,
+      "loss": 4.7545,
+      "step": 3304
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00042271228359439406,
+      "loss": 5.0942,
+      "step": 3305
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00042250618301731247,
+      "loss": 4.8582,
+      "step": 3306
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004223000824402308,
+      "loss": 4.8618,
+      "step": 3307
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00042209398186314923,
+      "loss": 4.9522,
+      "step": 3308
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0004218878812860676,
+      "loss": 4.8405,
+      "step": 3309
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000421681780708986,
+      "loss": 4.9969,
+      "step": 3310
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0004214756801319044,
+      "loss": 4.9493,
+      "step": 3311
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00042126957955482275,
+      "loss": 4.5735,
+      "step": 3312
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00042106347897774115,
+      "loss": 4.7574,
+      "step": 3313
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0004208573784006595,
+      "loss": 5.1326,
+      "step": 3314
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0004206512778235779,
+      "loss": 4.8311,
+      "step": 3315
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00042044517724649627,
+      "loss": 5.1197,
+      "step": 3316
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00042023907666941467,
+      "loss": 4.7596,
+      "step": 3317
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000420032976092333,
+      "loss": 4.7597,
+      "step": 3318
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00041982687551525143,
+      "loss": 4.6521,
+      "step": 3319
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00041962077493816984,
+      "loss": 4.7082,
+      "step": 3320
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0004194146743610882,
+      "loss": 4.7453,
+      "step": 3321
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0004192085737840066,
+      "loss": 4.9063,
+      "step": 3322
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00041900247320692495,
+      "loss": 4.9401,
+      "step": 3323
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00041879637262984336,
+      "loss": 4.6492,
+      "step": 3324
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0004185902720527617,
+      "loss": 4.5815,
+      "step": 3325
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0004183841714756801,
+      "loss": 5.5013,
+      "step": 3326
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0004181780708985986,
+      "loss": 5.1528,
+      "step": 3327
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00041797197032151693,
+      "loss": 4.7883,
+      "step": 3328
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00041776586974443534,
+      "loss": 4.9807,
+      "step": 3329
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004175597691673537,
+      "loss": 4.805,
+      "step": 3330
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004173536685902721,
+      "loss": 4.8616,
+      "step": 3331
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00041714756801319045,
+      "loss": 5.5066,
+      "step": 3332
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00041694146743610885,
+      "loss": 5.0608,
+      "step": 3333
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004167353668590272,
+      "loss": 4.4865,
+      "step": 3334
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004165292662819456,
+      "loss": 4.8919,
+      "step": 3335
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.000416323165704864,
+      "loss": 5.1962,
+      "step": 3336
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00041611706512778237,
+      "loss": 5.4543,
+      "step": 3337
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004159109645507008,
+      "loss": 4.9157,
+      "step": 3338
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00041570486397361913,
+      "loss": 4.5856,
+      "step": 3339
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00041549876339653754,
+      "loss": 4.8604,
+      "step": 3340
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004152926628194559,
+      "loss": 4.7516,
+      "step": 3341
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004150865622423743,
+      "loss": 5.0421,
+      "step": 3342
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0004148804616652927,
+      "loss": 5.024,
+      "step": 3343
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00041467436108821106,
+      "loss": 4.6225,
+      "step": 3344
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041446826051112946,
+      "loss": 5.5717,
+      "step": 3345
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0004142621599340478,
+      "loss": 4.6033,
+      "step": 3346
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0004140560593569662,
+      "loss": 4.5369,
+      "step": 3347
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0004138499587798846,
+      "loss": 4.7996,
+      "step": 3348
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000413643858202803,
+      "loss": 4.7199,
+      "step": 3349
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041343775762572133,
+      "loss": 4.8213,
+      "step": 3350
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041323165704863974,
+      "loss": 5.0071,
+      "step": 3351
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041302555647155815,
+      "loss": 4.8941,
+      "step": 3352
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0004128194558944765,
+      "loss": 4.7448,
+      "step": 3353
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0004126133553173949,
+      "loss": 5.0564,
+      "step": 3354
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041240725474031326,
+      "loss": 4.9331,
+      "step": 3355
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041220115416323167,
+      "loss": 4.5021,
+      "step": 3356
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041199505358615,
+      "loss": 5.2327,
+      "step": 3357
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0004117889530090684,
+      "loss": 4.5525,
+      "step": 3358
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041158285243198683,
+      "loss": 5.2163,
+      "step": 3359
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0004113767518549052,
+      "loss": 4.9207,
+      "step": 3360
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0004111706512778236,
+      "loss": 5.09,
+      "step": 3361
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00041096455070074194,
+      "loss": 5.0054,
+      "step": 3362
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00041075845012366035,
+      "loss": 4.8642,
+      "step": 3363
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0004105523495465787,
+      "loss": 4.6579,
+      "step": 3364
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0004103462489694971,
+      "loss": 5.4372,
+      "step": 3365
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00041014014839241546,
+      "loss": 4.9317,
+      "step": 3366
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00040993404781533387,
+      "loss": 5.0501,
+      "step": 3367
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0004097279472382523,
+      "loss": 5.1085,
+      "step": 3368
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00040952184666117063,
+      "loss": 5.0611,
+      "step": 3369
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00040931574608408903,
+      "loss": 4.8036,
+      "step": 3370
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0004091096455070074,
+      "loss": 5.104,
+      "step": 3371
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0004089035449299258,
+      "loss": 4.989,
+      "step": 3372
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00040869744435284415,
+      "loss": 4.7012,
+      "step": 3373
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0004084913437757626,
+      "loss": 5.0191,
+      "step": 3374
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.000408285243198681,
+      "loss": 5.4758,
+      "step": 3375
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00040807914262159937,
+      "loss": 4.7394,
+      "step": 3376
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00040787304204451777,
+      "loss": 4.7686,
+      "step": 3377
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0004076669414674361,
+      "loss": 4.6706,
+      "step": 3378
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00040746084089035453,
+      "loss": 4.6057,
+      "step": 3379
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0004072547403132729,
+      "loss": 4.5639,
+      "step": 3380
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0004070486397361913,
+      "loss": 6.4721,
+      "step": 3381
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0004068425391591097,
+      "loss": 4.9144,
+      "step": 3382
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00040663643858202805,
+      "loss": 5.0689,
+      "step": 3383
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00040643033800494646,
+      "loss": 4.5607,
+      "step": 3384
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0004062242374278648,
+      "loss": 4.8364,
+      "step": 3385
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0004060181368507832,
+      "loss": 4.8865,
+      "step": 3386
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00040581203627370157,
+      "loss": 4.9892,
+      "step": 3387
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00040560593569662,
+      "loss": 5.3338,
+      "step": 3388
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00040539983511953833,
+      "loss": 4.8478,
+      "step": 3389
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00040519373454245673,
+      "loss": 5.2041,
+      "step": 3390
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00040498763396537514,
+      "loss": 5.0093,
+      "step": 3391
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0004047815333882935,
+      "loss": 4.7164,
+      "step": 3392
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0004045754328112119,
+      "loss": 4.7543,
+      "step": 3393
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00040436933223413025,
+      "loss": 5.1934,
+      "step": 3394
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00040416323165704866,
+      "loss": 4.8449,
+      "step": 3395
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000403957131079967,
+      "loss": 5.1075,
+      "step": 3396
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0004037510305028854,
+      "loss": 4.9419,
+      "step": 3397
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0004035449299258038,
+      "loss": 5.0173,
+      "step": 3398
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0004033388293487222,
+      "loss": 4.8628,
+      "step": 3399
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0004031327287716406,
+      "loss": 4.8185,
+      "step": 3400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00040292662819455894,
+      "loss": 4.7231,
+      "step": 3401
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00040272052761747734,
+      "loss": 4.6144,
+      "step": 3402
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0004025144270403957,
+      "loss": 5.0075,
+      "step": 3403
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0004023083264633141,
+      "loss": 4.8532,
+      "step": 3404
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00040210222588623246,
+      "loss": 4.8107,
+      "step": 3405
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00040189612530915086,
+      "loss": 4.7789,
+      "step": 3406
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00040169002473206927,
+      "loss": 4.6945,
+      "step": 3407
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0004014839241549876,
+      "loss": 5.0356,
+      "step": 3408
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00040127782357790603,
+      "loss": 5.3554,
+      "step": 3409
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0004010717230008244,
+      "loss": 5.075,
+      "step": 3410
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0004008656224237428,
+      "loss": 4.8515,
+      "step": 3411
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00040065952184666114,
+      "loss": 4.9001,
+      "step": 3412
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00040045342126957955,
+      "loss": 4.6159,
+      "step": 3413
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00040024732069249795,
+      "loss": 4.7534,
+      "step": 3414
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0004000412201154163,
+      "loss": 4.8858,
+      "step": 3415
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0003998351195383347,
+      "loss": 4.7666,
+      "step": 3416
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039962901896125306,
+      "loss": 4.7242,
+      "step": 3417
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039942291838417147,
+      "loss": 4.6187,
+      "step": 3418
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003992168178070898,
+      "loss": 4.6584,
+      "step": 3419
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003990107172300083,
+      "loss": 4.6497,
+      "step": 3420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039880461665292664,
+      "loss": 5.1402,
+      "step": 3421
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039859851607584504,
+      "loss": 4.6371,
+      "step": 3422
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039839241549876345,
+      "loss": 5.1204,
+      "step": 3423
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003981863149216818,
+      "loss": 4.805,
+      "step": 3424
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003979802143446002,
+      "loss": 4.6992,
+      "step": 3425
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039777411376751856,
+      "loss": 4.6537,
+      "step": 3426
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039756801319043697,
+      "loss": 4.4712,
+      "step": 3427
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003973619126133553,
+      "loss": 4.8645,
+      "step": 3428
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039715581203627373,
+      "loss": 4.5011,
+      "step": 3429
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039694971145919213,
+      "loss": 5.5401,
+      "step": 3430
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003967436108821105,
+      "loss": 4.5239,
+      "step": 3431
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0003965375103050289,
+      "loss": 5.3311,
+      "step": 3432
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039633140972794725,
+      "loss": 4.3649,
+      "step": 3433
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00039612530915086565,
+      "loss": 4.8209,
+      "step": 3434
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.000395919208573784,
+      "loss": 4.9797,
+      "step": 3435
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003957131079967024,
+      "loss": 5.0191,
+      "step": 3436
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039550700741962076,
+      "loss": 4.7981,
+      "step": 3437
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039530090684253917,
+      "loss": 4.8585,
+      "step": 3438
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003950948062654576,
+      "loss": 4.7501,
+      "step": 3439
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039488870568837593,
+      "loss": 4.8708,
+      "step": 3440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039468260511129434,
+      "loss": 5.3529,
+      "step": 3441
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003944765045342127,
+      "loss": 5.1112,
+      "step": 3442
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003942704039571311,
+      "loss": 5.0601,
+      "step": 3443
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039406430338004945,
+      "loss": 4.8304,
+      "step": 3444
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039385820280296786,
+      "loss": 4.6868,
+      "step": 3445
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039365210222588626,
+      "loss": 4.9056,
+      "step": 3446
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003934460016488046,
+      "loss": 5.0938,
+      "step": 3447
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.000393239901071723,
+      "loss": 4.6568,
+      "step": 3448
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003930338004946414,
+      "loss": 4.7634,
+      "step": 3449
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0003928276999175598,
+      "loss": 5.1053,
+      "step": 3450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039262159934047813,
+      "loss": 4.8494,
+      "step": 3451
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00039241549876339654,
+      "loss": 4.8401,
+      "step": 3452
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003922093981863149,
+      "loss": 4.4674,
+      "step": 3453
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003920032976092333,
+      "loss": 4.7446,
+      "step": 3454
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003917971970321517,
+      "loss": 4.6724,
+      "step": 3455
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00039159109645507006,
+      "loss": 4.5054,
+      "step": 3456
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00039138499587798846,
+      "loss": 4.9614,
+      "step": 3457
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003911788953009068,
+      "loss": 4.9843,
+      "step": 3458
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003909727947238252,
+      "loss": 4.8049,
+      "step": 3459
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003907666941467436,
+      "loss": 4.6871,
+      "step": 3460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000390560593569662,
+      "loss": 4.6315,
+      "step": 3461
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003903544929925804,
+      "loss": 4.4,
+      "step": 3462
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00039014839241549874,
+      "loss": 5.0025,
+      "step": 3463
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00038994229183841715,
+      "loss": 4.6336,
+      "step": 3464
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003897361912613355,
+      "loss": 4.8333,
+      "step": 3465
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003895300906842539,
+      "loss": 5.0603,
+      "step": 3466
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003893239901071723,
+      "loss": 4.7075,
+      "step": 3467
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003891178895300907,
+      "loss": 4.3838,
+      "step": 3468
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0003889117889530091,
+      "loss": 4.5307,
+      "step": 3469
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003887056883759275,
+      "loss": 5.2701,
+      "step": 3470
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003884995877988459,
+      "loss": 4.5854,
+      "step": 3471
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00038829348722176424,
+      "loss": 4.7446,
+      "step": 3472
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00038808738664468265,
+      "loss": 5.5952,
+      "step": 3473
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000387881286067601,
+      "loss": 5.038,
+      "step": 3474
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003876751854905194,
+      "loss": 4.5915,
+      "step": 3475
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00038746908491343776,
+      "loss": 4.6189,
+      "step": 3476
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00038726298433635616,
+      "loss": 4.9148,
+      "step": 3477
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00038705688375927457,
+      "loss": 4.8993,
+      "step": 3478
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003868507831821929,
+      "loss": 5.1348,
+      "step": 3479
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00038664468260511133,
+      "loss": 4.7693,
+      "step": 3480
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003864385820280297,
+      "loss": 5.0197,
+      "step": 3481
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003862324814509481,
+      "loss": 4.5848,
+      "step": 3482
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00038602638087386644,
+      "loss": 4.9586,
+      "step": 3483
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00038582028029678485,
+      "loss": 4.7601,
+      "step": 3484
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00038561417971970326,
+      "loss": 5.0373,
+      "step": 3485
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0003854080791426216,
+      "loss": 4.6697,
+      "step": 3486
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00038520197856554,
+      "loss": 5.1292,
+      "step": 3487
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00038499587798845837,
+      "loss": 4.7371,
+      "step": 3488
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003847897774113768,
+      "loss": 4.8871,
+      "step": 3489
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003845836768342951,
+      "loss": 5.2292,
+      "step": 3490
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00038437757625721353,
+      "loss": 4.7833,
+      "step": 3491
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003841714756801319,
+      "loss": 5.0514,
+      "step": 3492
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003839653751030503,
+      "loss": 4.9005,
+      "step": 3493
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003837592745259687,
+      "loss": 4.8138,
+      "step": 3494
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00038355317394888705,
+      "loss": 4.8342,
+      "step": 3495
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00038334707337180546,
+      "loss": 5.0135,
+      "step": 3496
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003831409727947238,
+      "loss": 4.8047,
+      "step": 3497
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003829348722176422,
+      "loss": 4.7071,
+      "step": 3498
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00038272877164056057,
+      "loss": 4.8052,
+      "step": 3499
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.000382522671063479,
+      "loss": 4.6539,
+      "step": 3500
+    },
+    {
+      "epoch": 1.96,
+      "eval_loss": 5.350718975067139,
+      "eval_runtime": 1246.814,
+      "eval_samples_per_second": 2.119,
+      "eval_steps_per_second": 0.265,
+      "eval_wer": 1.960591035303451,
+      "step": 3500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003823165704863974,
+      "loss": 5.0306,
+      "step": 3501
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00038211046990931574,
+      "loss": 5.6189,
+      "step": 3502
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00038190436933223414,
+      "loss": 4.9781,
+      "step": 3503
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003816982687551525,
+      "loss": 4.4523,
+      "step": 3504
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0003814921681780709,
+      "loss": 4.4646,
+      "step": 3505
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00038128606760098925,
+      "loss": 4.9223,
+      "step": 3506
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00038107996702390766,
+      "loss": 5.0619,
+      "step": 3507
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.000380873866446826,
+      "loss": 4.8921,
+      "step": 3508
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003806677658697444,
+      "loss": 5.0318,
+      "step": 3509
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003804616652926628,
+      "loss": 4.9641,
+      "step": 3510
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003802555647155812,
+      "loss": 4.3759,
+      "step": 3511
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003800494641384996,
+      "loss": 4.8401,
+      "step": 3512
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00037984336356141794,
+      "loss": 4.7665,
+      "step": 3513
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003796372629843364,
+      "loss": 4.5638,
+      "step": 3514
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00037943116240725475,
+      "loss": 4.9765,
+      "step": 3515
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00037922506183017316,
+      "loss": 4.6742,
+      "step": 3516
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00037901896125309157,
+      "loss": 4.5614,
+      "step": 3517
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003788128606760099,
+      "loss": 4.933,
+      "step": 3518
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003786067600989283,
+      "loss": 4.6151,
+      "step": 3519
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003784006595218467,
+      "loss": 4.5681,
+      "step": 3520
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0003781945589447651,
+      "loss": 4.6579,
+      "step": 3521
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00037798845836768344,
+      "loss": 5.2545,
+      "step": 3522
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00037778235779060184,
+      "loss": 4.8947,
+      "step": 3523
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003775762572135202,
+      "loss": 5.0788,
+      "step": 3524
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003773701566364386,
+      "loss": 4.9018,
+      "step": 3525
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000377164056059357,
+      "loss": 5.1074,
+      "step": 3526
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00037695795548227536,
+      "loss": 4.903,
+      "step": 3527
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00037675185490519377,
+      "loss": 5.0159,
+      "step": 3528
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003765457543281121,
+      "loss": 4.7431,
+      "step": 3529
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003763396537510305,
+      "loss": 4.7598,
+      "step": 3530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003761335531739489,
+      "loss": 4.7656,
+      "step": 3531
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003759274525968673,
+      "loss": 4.7386,
+      "step": 3532
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003757213520197857,
+      "loss": 4.4311,
+      "step": 3533
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00037551525144270405,
+      "loss": 4.3768,
+      "step": 3534
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00037530915086562245,
+      "loss": 4.7547,
+      "step": 3535
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003751030502885408,
+      "loss": 4.8379,
+      "step": 3536
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003748969497114592,
+      "loss": 4.8409,
+      "step": 3537
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00037469084913437756,
+      "loss": 4.6882,
+      "step": 3538
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00037448474855729597,
+      "loss": 4.6677,
+      "step": 3539
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0003742786479802143,
+      "loss": 5.029,
+      "step": 3540
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00037407254740313273,
+      "loss": 5.1779,
+      "step": 3541
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00037386644682605114,
+      "loss": 4.6143,
+      "step": 3542
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0003736603462489695,
+      "loss": 5.2509,
+      "step": 3543
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0003734542456718879,
+      "loss": 4.6182,
+      "step": 3544
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00037324814509480625,
+      "loss": 4.5915,
+      "step": 3545
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00037304204451772465,
+      "loss": 4.7952,
+      "step": 3546
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.000372835943940643,
+      "loss": 4.9721,
+      "step": 3547
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0003726298433635614,
+      "loss": 4.9416,
+      "step": 3548
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0003724237427864798,
+      "loss": 4.6562,
+      "step": 3549
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00037221764220939817,
+      "loss": 4.7252,
+      "step": 3550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0003720115416323166,
+      "loss": 4.9539,
+      "step": 3551
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00037180544105523493,
+      "loss": 4.6379,
+      "step": 3552
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00037159934047815334,
+      "loss": 4.7811,
+      "step": 3553
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0003713932399010717,
+      "loss": 4.8201,
+      "step": 3554
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0003711871393239901,
+      "loss": 5.055,
+      "step": 3555
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00037098103874690845,
+      "loss": 5.5095,
+      "step": 3556
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00037077493816982686,
+      "loss": 4.7254,
+      "step": 3557
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00037056883759274526,
+      "loss": 4.8843,
+      "step": 3558
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0003703627370156636,
+      "loss": 4.9373,
+      "step": 3559
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003701566364385821,
+      "loss": 4.4665,
+      "step": 3560
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00036995053586150043,
+      "loss": 4.6459,
+      "step": 3561
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00036974443528441884,
+      "loss": 4.9304,
+      "step": 3562
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003695383347073372,
+      "loss": 4.9601,
+      "step": 3563
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003693322341302556,
+      "loss": 5.1502,
+      "step": 3564
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.000369126133553174,
+      "loss": 4.5378,
+      "step": 3565
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00036892003297609235,
+      "loss": 5.1122,
+      "step": 3566
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00036871393239901076,
+      "loss": 5.3363,
+      "step": 3567
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003685078318219291,
+      "loss": 3.7643,
+      "step": 3568
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003683017312448475,
+      "loss": 4.3605,
+      "step": 3569
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003680956306677659,
+      "loss": 4.4018,
+      "step": 3570
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003678895300906843,
+      "loss": 4.5606,
+      "step": 3571
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00036768342951360263,
+      "loss": 4.4119,
+      "step": 3572
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00036747732893652104,
+      "loss": 4.5348,
+      "step": 3573
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00036727122835943945,
+      "loss": 4.0751,
+      "step": 3574
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003670651277823578,
+      "loss": 4.3857,
+      "step": 3575
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0003668590272052762,
+      "loss": 4.2903,
+      "step": 3576
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00036665292662819456,
+      "loss": 4.4611,
+      "step": 3577
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00036644682605111296,
+      "loss": 4.2138,
+      "step": 3578
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003662407254740313,
+      "loss": 4.3851,
+      "step": 3579
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003660346248969497,
+      "loss": 4.3338,
+      "step": 3580
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00036582852431986813,
+      "loss": 4.4279,
+      "step": 3581
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003656224237427865,
+      "loss": 4.662,
+      "step": 3582
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003654163231657049,
+      "loss": 4.4004,
+      "step": 3583
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00036521022258862324,
+      "loss": 4.6863,
+      "step": 3584
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00036500412201154165,
+      "loss": 4.4686,
+      "step": 3585
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00036479802143446,
+      "loss": 4.331,
+      "step": 3586
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003645919208573784,
+      "loss": 4.4612,
+      "step": 3587
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00036438582028029676,
+      "loss": 4.2718,
+      "step": 3588
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00036417971970321517,
+      "loss": 4.2433,
+      "step": 3589
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003639736191261336,
+      "loss": 4.0857,
+      "step": 3590
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003637675185490519,
+      "loss": 4.5854,
+      "step": 3591
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00036356141797197033,
+      "loss": 4.6846,
+      "step": 3592
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003633553173948887,
+      "loss": 4.1719,
+      "step": 3593
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0003631492168178071,
+      "loss": 4.2534,
+      "step": 3594
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00036294311624072544,
+      "loss": 4.1953,
+      "step": 3595
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00036273701566364385,
+      "loss": 4.2721,
+      "step": 3596
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00036253091508656226,
+      "loss": 4.2088,
+      "step": 3597
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003623248145094806,
+      "loss": 4.4096,
+      "step": 3598
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.000362118713932399,
+      "loss": 4.4036,
+      "step": 3599
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00036191261335531737,
+      "loss": 4.2923,
+      "step": 3600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003617065127782358,
+      "loss": 4.2254,
+      "step": 3601
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00036150041220115413,
+      "loss": 4.0549,
+      "step": 3602
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00036129431162407253,
+      "loss": 3.994,
+      "step": 3603
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00036108821104699094,
+      "loss": 4.2491,
+      "step": 3604
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003608821104699093,
+      "loss": 4.4375,
+      "step": 3605
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003606760098928277,
+      "loss": 4.1345,
+      "step": 3606
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003604699093157461,
+      "loss": 4.0748,
+      "step": 3607
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003602638087386645,
+      "loss": 4.2856,
+      "step": 3608
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00036005770816158287,
+      "loss": 4.2659,
+      "step": 3609
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003598516075845013,
+      "loss": 4.3432,
+      "step": 3610
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0003596455070074196,
+      "loss": 3.9874,
+      "step": 3611
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00035943940643033803,
+      "loss": 4.0017,
+      "step": 3612
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00035923330585325644,
+      "loss": 3.8684,
+      "step": 3613
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003590272052761748,
+      "loss": 4.0896,
+      "step": 3614
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003588211046990932,
+      "loss": 3.4065,
+      "step": 3615
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00035861500412201155,
+      "loss": 3.5706,
+      "step": 3616
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00035840890354492996,
+      "loss": 3.7421,
+      "step": 3617
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003582028029678483,
+      "loss": 4.1022,
+      "step": 3618
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003579967023907667,
+      "loss": 4.4687,
+      "step": 3619
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003577906018136851,
+      "loss": 4.1813,
+      "step": 3620
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003575845012366035,
+      "loss": 4.4015,
+      "step": 3621
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003573784006595219,
+      "loss": 4.4738,
+      "step": 3622
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00035717230008244024,
+      "loss": 4.2639,
+      "step": 3623
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00035696619950535864,
+      "loss": 4.3712,
+      "step": 3624
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.000356760098928277,
+      "loss": 4.3999,
+      "step": 3625
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003565539983511954,
+      "loss": 4.5484,
+      "step": 3626
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00035634789777411375,
+      "loss": 4.405,
+      "step": 3627
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00035614179719703216,
+      "loss": 4.6665,
+      "step": 3628
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00035593569661995057,
+      "loss": 4.5232,
+      "step": 3629
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0003557295960428689,
+      "loss": 4.1397,
+      "step": 3630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003555234954657873,
+      "loss": 4.2026,
+      "step": 3631
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003553173948887057,
+      "loss": 4.2076,
+      "step": 3632
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003551112943116241,
+      "loss": 4.8621,
+      "step": 3633
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00035490519373454244,
+      "loss": 4.3434,
+      "step": 3634
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00035469909315746084,
+      "loss": 4.4534,
+      "step": 3635
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00035449299258037925,
+      "loss": 4.3467,
+      "step": 3636
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003542868920032976,
+      "loss": 4.5217,
+      "step": 3637
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.000354080791426216,
+      "loss": 4.3418,
+      "step": 3638
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00035387469084913436,
+      "loss": 4.1879,
+      "step": 3639
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00035366859027205277,
+      "loss": 4.2386,
+      "step": 3640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003534624896949711,
+      "loss": 4.2325,
+      "step": 3641
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00035325638911788953,
+      "loss": 4.0876,
+      "step": 3642
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003530502885408079,
+      "loss": 4.3226,
+      "step": 3643
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003528441879637263,
+      "loss": 4.3533,
+      "step": 3644
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003526380873866447,
+      "loss": 4.0237,
+      "step": 3645
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00035243198680956305,
+      "loss": 4.1715,
+      "step": 3646
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00035222588623248145,
+      "loss": 4.3486,
+      "step": 3647
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0003520197856553998,
+      "loss": 4.2525,
+      "step": 3648
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003518136850783182,
+      "loss": 4.3741,
+      "step": 3649
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00035160758450123657,
+      "loss": 4.0705,
+      "step": 3650
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00035140148392415497,
+      "loss": 4.6507,
+      "step": 3651
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003511953833470734,
+      "loss": 4.2882,
+      "step": 3652
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00035098928276999173,
+      "loss": 4.4146,
+      "step": 3653
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003507831821929102,
+      "loss": 3.8574,
+      "step": 3654
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00035057708161582854,
+      "loss": 4.2135,
+      "step": 3655
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00035037098103874695,
+      "loss": 4.1638,
+      "step": 3656
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003501648804616653,
+      "loss": 4.3172,
+      "step": 3657
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003499587798845837,
+      "loss": 4.0007,
+      "step": 3658
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00034975267930750206,
+      "loss": 4.1149,
+      "step": 3659
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00034954657873042047,
+      "loss": 4.3865,
+      "step": 3660
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003493404781533389,
+      "loss": 4.3148,
+      "step": 3661
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00034913437757625723,
+      "loss": 4.0123,
+      "step": 3662
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00034892827699917564,
+      "loss": 3.9729,
+      "step": 3663
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.000348722176422094,
+      "loss": 3.969,
+      "step": 3664
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0003485160758450124,
+      "loss": 4.3417,
+      "step": 3665
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00034830997526793075,
+      "loss": 3.9341,
+      "step": 3666
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00034810387469084915,
+      "loss": 3.9295,
+      "step": 3667
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00034789777411376756,
+      "loss": 3.6904,
+      "step": 3668
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003476916735366859,
+      "loss": 4.6604,
+      "step": 3669
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003474855729596043,
+      "loss": 4.8435,
+      "step": 3670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00034727947238252267,
+      "loss": 4.2181,
+      "step": 3671
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003470733718054411,
+      "loss": 4.326,
+      "step": 3672
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00034686727122835943,
+      "loss": 4.302,
+      "step": 3673
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00034666117065127784,
+      "loss": 4.2657,
+      "step": 3674
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003464550700741962,
+      "loss": 4.8252,
+      "step": 3675
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003462489694971146,
+      "loss": 4.3396,
+      "step": 3676
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000346042868920033,
+      "loss": 4.585,
+      "step": 3677
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00034583676834295136,
+      "loss": 4.2891,
+      "step": 3678
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00034563066776586976,
+      "loss": 4.2514,
+      "step": 3679
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003454245671887881,
+      "loss": 4.3842,
+      "step": 3680
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003452184666117065,
+      "loss": 4.3541,
+      "step": 3681
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003450123660346249,
+      "loss": 4.2877,
+      "step": 3682
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0003448062654575433,
+      "loss": 4.337,
+      "step": 3683
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003446001648804617,
+      "loss": 4.4383,
+      "step": 3684
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00034439406430338004,
+      "loss": 4.6116,
+      "step": 3685
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00034418796372629845,
+      "loss": 4.5534,
+      "step": 3686
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003439818631492168,
+      "loss": 4.5576,
+      "step": 3687
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003437757625721352,
+      "loss": 4.5087,
+      "step": 3688
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00034356966199505356,
+      "loss": 4.2941,
+      "step": 3689
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00034336356141797197,
+      "loss": 4.5397,
+      "step": 3690
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003431574608408903,
+      "loss": 4.1435,
+      "step": 3691
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003429513602638087,
+      "loss": 4.4491,
+      "step": 3692
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00034274525968672713,
+      "loss": 4.0729,
+      "step": 3693
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003425391591096455,
+      "loss": 4.2249,
+      "step": 3694
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003423330585325639,
+      "loss": 4.3708,
+      "step": 3695
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00034212695795548224,
+      "loss": 4.1194,
+      "step": 3696
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00034192085737840065,
+      "loss": 4.4926,
+      "step": 3697
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000341714756801319,
+      "loss": 4.1033,
+      "step": 3698
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003415086562242374,
+      "loss": 4.2112,
+      "step": 3699
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00034130255564715587,
+      "loss": 4.304,
+      "step": 3700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0003410964550700742,
+      "loss": 4.6552,
+      "step": 3701
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00034089035449299263,
+      "loss": 4.4121,
+      "step": 3702
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000340684253915911,
+      "loss": 4.2547,
+      "step": 3703
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003404781533388294,
+      "loss": 4.4291,
+      "step": 3704
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00034027205276174774,
+      "loss": 4.2755,
+      "step": 3705
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00034006595218466615,
+      "loss": 4.53,
+      "step": 3706
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00033985985160758455,
+      "loss": 4.2598,
+      "step": 3707
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003396537510305029,
+      "loss": 4.1691,
+      "step": 3708
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003394476504534213,
+      "loss": 4.4356,
+      "step": 3709
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00033924154987633967,
+      "loss": 4.1657,
+      "step": 3710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00033903544929925807,
+      "loss": 4.3095,
+      "step": 3711
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003388293487221764,
+      "loss": 4.9467,
+      "step": 3712
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00033862324814509483,
+      "loss": 4.2679,
+      "step": 3713
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003384171475680132,
+      "loss": 4.3145,
+      "step": 3714
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003382110469909316,
+      "loss": 4.8211,
+      "step": 3715
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00033800494641385,
+      "loss": 3.6923,
+      "step": 3716
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00033779884583676835,
+      "loss": 4.068,
+      "step": 3717
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00033759274525968676,
+      "loss": 3.5284,
+      "step": 3718
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0003373866446826051,
+      "loss": 4.7115,
+      "step": 3719
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003371805441055235,
+      "loss": 4.8537,
+      "step": 3720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00033697444352844187,
+      "loss": 4.1185,
+      "step": 3721
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003367683429513603,
+      "loss": 4.2886,
+      "step": 3722
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003365622423742787,
+      "loss": 5.039,
+      "step": 3723
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00033635614179719703,
+      "loss": 4.264,
+      "step": 3724
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00033615004122011544,
+      "loss": 4.3003,
+      "step": 3725
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003359439406430338,
+      "loss": 4.2089,
+      "step": 3726
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003357378400659522,
+      "loss": 4.1266,
+      "step": 3727
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00033553173948887055,
+      "loss": 4.249,
+      "step": 3728
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00033532563891178896,
+      "loss": 4.4275,
+      "step": 3729
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003351195383347073,
+      "loss": 4.4754,
+      "step": 3730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003349134377576257,
+      "loss": 4.0129,
+      "step": 3731
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003347073371805441,
+      "loss": 4.3328,
+      "step": 3732
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003345012366034625,
+      "loss": 4.3133,
+      "step": 3733
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0003342951360263809,
+      "loss": 4.6834,
+      "step": 3734
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00033408903544929924,
+      "loss": 4.0161,
+      "step": 3735
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00033388293487221764,
+      "loss": 4.4901,
+      "step": 3736
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.000333676834295136,
+      "loss": 4.2939,
+      "step": 3737
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003334707337180544,
+      "loss": 4.3622,
+      "step": 3738
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003332646331409728,
+      "loss": 4.3752,
+      "step": 3739
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00033305853256389116,
+      "loss": 4.4574,
+      "step": 3740
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00033285243198680957,
+      "loss": 4.2626,
+      "step": 3741
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003326463314097279,
+      "loss": 4.3466,
+      "step": 3742
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00033244023083264633,
+      "loss": 4.2333,
+      "step": 3743
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003322341302555647,
+      "loss": 4.4966,
+      "step": 3744
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003320280296784831,
+      "loss": 4.3876,
+      "step": 3745
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00033182192910140144,
+      "loss": 4.4672,
+      "step": 3746
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003316158285243199,
+      "loss": 3.9761,
+      "step": 3747
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003314097279472383,
+      "loss": 4.0863,
+      "step": 3748
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00033120362737015666,
+      "loss": 4.1063,
+      "step": 3749
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00033099752679307507,
+      "loss": 4.4057,
+      "step": 3750
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003307914262159934,
+      "loss": 4.2496,
+      "step": 3751
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003305853256389118,
+      "loss": 4.2922,
+      "step": 3752
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003303792250618302,
+      "loss": 4.1843,
+      "step": 3753
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0003301731244847486,
+      "loss": 3.9541,
+      "step": 3754
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000329967023907667,
+      "loss": 4.4302,
+      "step": 3755
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00032976092333058534,
+      "loss": 4.1149,
+      "step": 3756
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00032955482275350375,
+      "loss": 4.3257,
+      "step": 3757
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003293487221764221,
+      "loss": 4.3969,
+      "step": 3758
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003291426215993405,
+      "loss": 4.5217,
+      "step": 3759
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00032893652102225886,
+      "loss": 4.7026,
+      "step": 3760
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00032873042044517727,
+      "loss": 4.0604,
+      "step": 3761
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003285243198680956,
+      "loss": 4.1155,
+      "step": 3762
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00032831821929101403,
+      "loss": 4.2195,
+      "step": 3763
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00032811211871393243,
+      "loss": 3.9685,
+      "step": 3764
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003279060181368508,
+      "loss": 3.965,
+      "step": 3765
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003276999175597692,
+      "loss": 4.0773,
+      "step": 3766
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00032749381698268755,
+      "loss": 4.4035,
+      "step": 3767
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00032728771640560595,
+      "loss": 3.7457,
+      "step": 3768
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003270816158285243,
+      "loss": 4.54,
+      "step": 3769
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003268755152514427,
+      "loss": 4.1146,
+      "step": 3770
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003266694146743611,
+      "loss": 4.4491,
+      "step": 3771
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00032646331409727947,
+      "loss": 4.5075,
+      "step": 3772
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0003262572135201979,
+      "loss": 4.2792,
+      "step": 3773
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00032605111294311623,
+      "loss": 4.2084,
+      "step": 3774
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00032584501236603464,
+      "loss": 4.2807,
+      "step": 3775
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.000325638911788953,
+      "loss": 4.4187,
+      "step": 3776
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003254328112118714,
+      "loss": 4.4392,
+      "step": 3777
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00032522671063478975,
+      "loss": 4.4915,
+      "step": 3778
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00032502061005770816,
+      "loss": 4.1347,
+      "step": 3779
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00032481450948062656,
+      "loss": 4.1635,
+      "step": 3780
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003246084089035449,
+      "loss": 4.3304,
+      "step": 3781
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003244023083264633,
+      "loss": 4.856,
+      "step": 3782
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003241962077493817,
+      "loss": 3.8823,
+      "step": 3783
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003239901071723001,
+      "loss": 4.2622,
+      "step": 3784
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00032378400659521843,
+      "loss": 4.4088,
+      "step": 3785
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00032357790601813684,
+      "loss": 4.1953,
+      "step": 3786
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00032337180544105525,
+      "loss": 4.5575,
+      "step": 3787
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0003231657048639736,
+      "loss": 4.3683,
+      "step": 3788
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.000322959604286892,
+      "loss": 4.176,
+      "step": 3789
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00032275350370981036,
+      "loss": 4.6127,
+      "step": 3790
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00032254740313272876,
+      "loss": 4.4243,
+      "step": 3791
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003223413025556471,
+      "loss": 4.6342,
+      "step": 3792
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003221352019785655,
+      "loss": 4.2625,
+      "step": 3793
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00032192910140148393,
+      "loss": 4.2286,
+      "step": 3794
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00032172300082440234,
+      "loss": 4.1858,
+      "step": 3795
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00032151690024732074,
+      "loss": 4.7901,
+      "step": 3796
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003213107996702391,
+      "loss": 4.3982,
+      "step": 3797
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003211046990931575,
+      "loss": 4.3167,
+      "step": 3798
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00032089859851607586,
+      "loss": 4.261,
+      "step": 3799
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00032069249793899426,
+      "loss": 4.362,
+      "step": 3800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003204863973619126,
+      "loss": 4.0453,
+      "step": 3801
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000320280296784831,
+      "loss": 4.4057,
+      "step": 3802
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00032007419620774943,
+      "loss": 4.1651,
+      "step": 3803
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003198680956306678,
+      "loss": 4.2865,
+      "step": 3804
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003196619950535862,
+      "loss": 3.9848,
+      "step": 3805
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00031945589447650454,
+      "loss": 4.2811,
+      "step": 3806
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00031924979389942295,
+      "loss": 3.8514,
+      "step": 3807
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0003190436933223413,
+      "loss": 3.9445,
+      "step": 3808
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003188375927452597,
+      "loss": 4.0056,
+      "step": 3809
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00031863149216817806,
+      "loss": 4.1554,
+      "step": 3810
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00031842539159109646,
+      "loss": 4.1162,
+      "step": 3811
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00031821929101401487,
+      "loss": 3.5571,
+      "step": 3812
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003180131904369332,
+      "loss": 3.9999,
+      "step": 3813
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00031780708985985163,
+      "loss": 3.833,
+      "step": 3814
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00031760098928277,
+      "loss": 3.8088,
+      "step": 3815
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003173948887056884,
+      "loss": 3.632,
+      "step": 3816
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00031718878812860674,
+      "loss": 3.5225,
+      "step": 3817
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00031698268755152515,
+      "loss": 4.094,
+      "step": 3818
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00031677658697444356,
+      "loss": 4.6527,
+      "step": 3819
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003165704863973619,
+      "loss": 4.4415,
+      "step": 3820
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003163643858202803,
+      "loss": 4.5488,
+      "step": 3821
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00031615828524319867,
+      "loss": 4.2311,
+      "step": 3822
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003159521846661171,
+      "loss": 4.5498,
+      "step": 3823
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0003157460840890354,
+      "loss": 4.5728,
+      "step": 3824
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00031553998351195383,
+      "loss": 4.5634,
+      "step": 3825
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00031533388293487224,
+      "loss": 4.2564,
+      "step": 3826
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003151277823577906,
+      "loss": 4.9106,
+      "step": 3827
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.000314921681780709,
+      "loss": 4.4247,
+      "step": 3828
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00031471558120362735,
+      "loss": 4.3949,
+      "step": 3829
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00031450948062654576,
+      "loss": 4.2906,
+      "step": 3830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003143033800494641,
+      "loss": 4.6022,
+      "step": 3831
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003140972794723825,
+      "loss": 4.1676,
+      "step": 3832
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00031389117889530087,
+      "loss": 4.1554,
+      "step": 3833
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003136850783182193,
+      "loss": 4.5186,
+      "step": 3834
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003134789777411377,
+      "loss": 4.2841,
+      "step": 3835
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00031327287716405604,
+      "loss": 4.2017,
+      "step": 3836
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00031306677658697444,
+      "loss": 4.2012,
+      "step": 3837
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003128606760098928,
+      "loss": 4.3024,
+      "step": 3838
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003126545754328112,
+      "loss": 4.1568,
+      "step": 3839
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00031244847485572955,
+      "loss": 4.3771,
+      "step": 3840
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.000312242374278648,
+      "loss": 4.5044,
+      "step": 3841
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003120362737015664,
+      "loss": 4.2441,
+      "step": 3842
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003118301731244848,
+      "loss": 4.3266,
+      "step": 3843
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0003116240725474032,
+      "loss": 4.1094,
+      "step": 3844
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00031141797197032153,
+      "loss": 4.0582,
+      "step": 3845
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00031121187139323994,
+      "loss": 4.3133,
+      "step": 3846
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0003110057708161583,
+      "loss": 4.6793,
+      "step": 3847
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0003107996702390767,
+      "loss": 4.7592,
+      "step": 3848
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00031059356966199505,
+      "loss": 4.2604,
+      "step": 3849
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00031038746908491346,
+      "loss": 4.1127,
+      "step": 3850
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00031018136850783186,
+      "loss": 4.0515,
+      "step": 3851
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0003099752679307502,
+      "loss": 4.0861,
+      "step": 3852
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0003097691673536686,
+      "loss": 4.0447,
+      "step": 3853
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.000309563066776587,
+      "loss": 3.8888,
+      "step": 3854
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0003093569661995054,
+      "loss": 4.2354,
+      "step": 3855
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00030915086562242374,
+      "loss": 4.092,
+      "step": 3856
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00030894476504534214,
+      "loss": 4.1411,
+      "step": 3857
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00030873866446826055,
+      "loss": 4.2234,
+      "step": 3858
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0003085325638911789,
+      "loss": 4.6418,
+      "step": 3859
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0003083264633140973,
+      "loss": 4.2529,
+      "step": 3860
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00030812036273701566,
+      "loss": 4.1276,
+      "step": 3861
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00030791426215993407,
+      "loss": 3.697,
+      "step": 3862
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003077081615828524,
+      "loss": 3.6277,
+      "step": 3863
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003075020610057708,
+      "loss": 4.0292,
+      "step": 3864
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003072959604286892,
+      "loss": 3.9038,
+      "step": 3865
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003070898598516076,
+      "loss": 3.6629,
+      "step": 3866
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.000306883759274526,
+      "loss": 4.7387,
+      "step": 3867
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00030667765869744435,
+      "loss": 3.7893,
+      "step": 3868
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00030647155812036275,
+      "loss": 4.4118,
+      "step": 3869
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003062654575432811,
+      "loss": 4.5624,
+      "step": 3870
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003060593569661995,
+      "loss": 4.3192,
+      "step": 3871
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00030585325638911786,
+      "loss": 4.3452,
+      "step": 3872
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00030564715581203627,
+      "loss": 4.5882,
+      "step": 3873
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003054410552349547,
+      "loss": 4.5963,
+      "step": 3874
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00030523495465787303,
+      "loss": 4.1869,
+      "step": 3875
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00030502885408079144,
+      "loss": 4.3875,
+      "step": 3876
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003048227535037098,
+      "loss": 4.3676,
+      "step": 3877
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0003046166529266282,
+      "loss": 4.2629,
+      "step": 3878
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00030441055234954655,
+      "loss": 4.4547,
+      "step": 3879
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00030420445177246495,
+      "loss": 4.6379,
+      "step": 3880
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003039983511953833,
+      "loss": 4.2814,
+      "step": 3881
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003037922506183017,
+      "loss": 4.3674,
+      "step": 3882
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003035861500412201,
+      "loss": 4.3673,
+      "step": 3883
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00030338004946413847,
+      "loss": 4.5122,
+      "step": 3884
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003031739488870569,
+      "loss": 4.1023,
+      "step": 3885
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00030296784830997523,
+      "loss": 4.1202,
+      "step": 3886
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003027617477328937,
+      "loss": 4.0869,
+      "step": 3887
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00030255564715581205,
+      "loss": 4.1754,
+      "step": 3888
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00030234954657873045,
+      "loss": 4.0549,
+      "step": 3889
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00030214344600164886,
+      "loss": 4.424,
+      "step": 3890
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003019373454245672,
+      "loss": 4.1914,
+      "step": 3891
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003017312448474856,
+      "loss": 4.2785,
+      "step": 3892
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00030152514427040397,
+      "loss": 4.1787,
+      "step": 3893
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003013190436933224,
+      "loss": 4.1266,
+      "step": 3894
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00030111294311624073,
+      "loss": 4.0083,
+      "step": 3895
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00030090684253915914,
+      "loss": 4.266,
+      "step": 3896
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003007007419620775,
+      "loss": 4.2184,
+      "step": 3897
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0003004946413849959,
+      "loss": 4.2778,
+      "step": 3898
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0003002885408079143,
+      "loss": 3.8303,
+      "step": 3899
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00030008244023083265,
+      "loss": 4.0113,
+      "step": 3900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00029987633965375106,
+      "loss": 4.1548,
+      "step": 3901
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002996702390766694,
+      "loss": 4.1023,
+      "step": 3902
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002994641384995878,
+      "loss": 4.3791,
+      "step": 3903
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00029925803792250617,
+      "loss": 4.3735,
+      "step": 3904
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002990519373454246,
+      "loss": 4.3744,
+      "step": 3905
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.000298845836768343,
+      "loss": 4.2447,
+      "step": 3906
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00029863973619126134,
+      "loss": 4.1391,
+      "step": 3907
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00029843363561417975,
+      "loss": 4.1146,
+      "step": 3908
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002982275350370981,
+      "loss": 4.4045,
+      "step": 3909
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002980214344600165,
+      "loss": 4.8299,
+      "step": 3910
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00029781533388293486,
+      "loss": 4.1358,
+      "step": 3911
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00029760923330585326,
+      "loss": 3.9755,
+      "step": 3912
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002974031327287716,
+      "loss": 3.9685,
+      "step": 3913
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00029719703215169,
+      "loss": 3.9762,
+      "step": 3914
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00029699093157460843,
+      "loss": 4.2899,
+      "step": 3915
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002967848309975268,
+      "loss": 3.6374,
+      "step": 3916
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002965787304204452,
+      "loss": 4.0514,
+      "step": 3917
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00029637262984336354,
+      "loss": 3.6694,
+      "step": 3918
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00029616652926628195,
+      "loss": 4.8307,
+      "step": 3919
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002959604286892003,
+      "loss": 4.364,
+      "step": 3920
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002957543281121187,
+      "loss": 4.1227,
+      "step": 3921
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002955482275350371,
+      "loss": 4.2025,
+      "step": 3922
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00029534212695795547,
+      "loss": 4.579,
+      "step": 3923
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002951360263808739,
+      "loss": 4.4036,
+      "step": 3924
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002949299258037922,
+      "loss": 4.2262,
+      "step": 3925
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00029472382522671063,
+      "loss": 4.2305,
+      "step": 3926
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.000294517724649629,
+      "loss": 4.3236,
+      "step": 3927
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002943116240725474,
+      "loss": 4.1951,
+      "step": 3928
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002941055234954658,
+      "loss": 4.3775,
+      "step": 3929
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00029389942291838415,
+      "loss": 4.2302,
+      "step": 3930
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00029369332234130256,
+      "loss": 4.2603,
+      "step": 3931
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002934872217642209,
+      "loss": 4.145,
+      "step": 3932
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00029328112118713937,
+      "loss": 4.2158,
+      "step": 3933
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002930750206100577,
+      "loss": 4.1811,
+      "step": 3934
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00029286892003297613,
+      "loss": 4.327,
+      "step": 3935
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002926628194558945,
+      "loss": 4.0478,
+      "step": 3936
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002924567188788129,
+      "loss": 4.4285,
+      "step": 3937
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002922506183017313,
+      "loss": 4.3073,
+      "step": 3938
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00029204451772464965,
+      "loss": 4.0413,
+      "step": 3939
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00029183841714756805,
+      "loss": 4.1638,
+      "step": 3940
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002916323165704864,
+      "loss": 4.1041,
+      "step": 3941
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002914262159934048,
+      "loss": 3.9455,
+      "step": 3942
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00029122011541632317,
+      "loss": 4.3891,
+      "step": 3943
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002910140148392416,
+      "loss": 4.1985,
+      "step": 3944
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00029080791426216,
+      "loss": 4.1345,
+      "step": 3945
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00029060181368507833,
+      "loss": 4.351,
+      "step": 3946
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00029039571310799674,
+      "loss": 4.1234,
+      "step": 3947
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002901896125309151,
+      "loss": 4.4742,
+      "step": 3948
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0002899835119538335,
+      "loss": 4.3027,
+      "step": 3949
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00028977741137675185,
+      "loss": 3.8949,
+      "step": 3950
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00028957131079967026,
+      "loss": 4.0097,
+      "step": 3951
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002893652102225886,
+      "loss": 4.334,
+      "step": 3952
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.000289159109645507,
+      "loss": 4.0044,
+      "step": 3953
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002889530090684254,
+      "loss": 3.9495,
+      "step": 3954
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002887469084913438,
+      "loss": 3.9834,
+      "step": 3955
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002885408079142622,
+      "loss": 4.1475,
+      "step": 3956
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00028833470733718053,
+      "loss": 4.0109,
+      "step": 3957
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00028812860676009894,
+      "loss": 3.9122,
+      "step": 3958
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002879225061830173,
+      "loss": 3.8312,
+      "step": 3959
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002877164056059357,
+      "loss": 4.2956,
+      "step": 3960
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002875103050288541,
+      "loss": 3.8791,
+      "step": 3961
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00028730420445177246,
+      "loss": 3.9732,
+      "step": 3962
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00028709810387469087,
+      "loss": 4.3162,
+      "step": 3963
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002868920032976092,
+      "loss": 4.3692,
+      "step": 3964
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002866859027205276,
+      "loss": 3.7563,
+      "step": 3965
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.000286479802143446,
+      "loss": 4.2954,
+      "step": 3966
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002862737015663644,
+      "loss": 5.4757,
+      "step": 3967
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00028606760098928274,
+      "loss": 3.7368,
+      "step": 3968
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00028586150041220114,
+      "loss": 4.6096,
+      "step": 3969
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00028565539983511955,
+      "loss": 4.5703,
+      "step": 3970
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002854492992580379,
+      "loss": 4.5912,
+      "step": 3971
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002852431986809563,
+      "loss": 4.1268,
+      "step": 3972
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00028503709810387466,
+      "loss": 4.634,
+      "step": 3973
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00028483099752679307,
+      "loss": 4.0544,
+      "step": 3974
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002846248969497114,
+      "loss": 4.429,
+      "step": 3975
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00028441879637262983,
+      "loss": 4.6517,
+      "step": 3976
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00028421269579554823,
+      "loss": 4.1454,
+      "step": 3977
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002840065952184666,
+      "loss": 4.1397,
+      "step": 3978
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.000283800494641385,
+      "loss": 4.6336,
+      "step": 3979
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002835943940643034,
+      "loss": 4.4107,
+      "step": 3980
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002833882934872218,
+      "loss": 3.9808,
+      "step": 3981
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00028318219291014016,
+      "loss": 4.108,
+      "step": 3982
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00028297609233305857,
+      "loss": 4.277,
+      "step": 3983
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002827699917559769,
+      "loss": 4.2215,
+      "step": 3984
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002825638911788953,
+      "loss": 4.2538,
+      "step": 3985
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00028235779060181373,
+      "loss": 4.1574,
+      "step": 3986
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002821516900247321,
+      "loss": 4.1647,
+      "step": 3987
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002819455894476505,
+      "loss": 4.2905,
+      "step": 3988
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00028173948887056884,
+      "loss": 4.0986,
+      "step": 3989
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00028153338829348725,
+      "loss": 4.4267,
+      "step": 3990
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002813272877164056,
+      "loss": 4.4692,
+      "step": 3991
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.000281121187139324,
+      "loss": 4.3102,
+      "step": 3992
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002809150865622424,
+      "loss": 4.0897,
+      "step": 3993
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00028070898598516077,
+      "loss": 4.3005,
+      "step": 3994
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002805028854080792,
+      "loss": 4.4165,
+      "step": 3995
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00028029678483099753,
+      "loss": 3.9828,
+      "step": 3996
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00028009068425391594,
+      "loss": 4.5387,
+      "step": 3997
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002798845836768343,
+      "loss": 4.4824,
+      "step": 3998
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002796784830997527,
+      "loss": 4.2889,
+      "step": 3999
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00027947238252267105,
+      "loss": 4.351,
+      "step": 4000
+    },
+    {
+      "epoch": 2.24,
+      "eval_loss": 5.417797565460205,
+      "eval_runtime": 1245.9851,
+      "eval_samples_per_second": 2.12,
+      "eval_steps_per_second": 0.266,
+      "eval_wer": 1.9677707259024197,
+      "step": 4000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00027926628194558945,
+      "loss": 4.2183,
+      "step": 4001
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00027906018136850786,
+      "loss": 4.4004,
+      "step": 4002
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002788540807914262,
+      "loss": 4.1819,
+      "step": 4003
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002786479802143446,
+      "loss": 4.009,
+      "step": 4004
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00027844187963726297,
+      "loss": 4.6786,
+      "step": 4005
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002782357790601814,
+      "loss": 4.2294,
+      "step": 4006
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00027802967848309973,
+      "loss": 4.1328,
+      "step": 4007
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00027782357790601814,
+      "loss": 4.3012,
+      "step": 4008
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00027761747732893654,
+      "loss": 4.2026,
+      "step": 4009
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002774113767518549,
+      "loss": 4.2979,
+      "step": 4010
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002772052761747733,
+      "loss": 4.4374,
+      "step": 4011
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00027699917559769166,
+      "loss": 3.9464,
+      "step": 4012
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00027679307502061006,
+      "loss": 3.714,
+      "step": 4013
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002765869744435284,
+      "loss": 3.8914,
+      "step": 4014
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002763808738664468,
+      "loss": 3.7032,
+      "step": 4015
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002761747732893652,
+      "loss": 4.1146,
+      "step": 4016
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002759686727122836,
+      "loss": 3.9557,
+      "step": 4017
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000275762572135202,
+      "loss": 4.0967,
+      "step": 4018
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00027555647155812034,
+      "loss": 4.5957,
+      "step": 4019
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00027535037098103875,
+      "loss": 4.3779,
+      "step": 4020
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002751442704039571,
+      "loss": 4.2485,
+      "step": 4021
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002749381698268755,
+      "loss": 4.2819,
+      "step": 4022
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00027473206924979386,
+      "loss": 4.1107,
+      "step": 4023
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00027452596867271227,
+      "loss": 4.3118,
+      "step": 4024
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00027431986809563067,
+      "loss": 4.8528,
+      "step": 4025
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000274113767518549,
+      "loss": 4.4267,
+      "step": 4026
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002739076669414675,
+      "loss": 4.6165,
+      "step": 4027
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00027370156636438584,
+      "loss": 4.4067,
+      "step": 4028
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00027349546578730424,
+      "loss": 4.2029,
+      "step": 4029
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002732893652102226,
+      "loss": 4.0953,
+      "step": 4030
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000273083264633141,
+      "loss": 4.1296,
+      "step": 4031
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00027287716405605936,
+      "loss": 4.2069,
+      "step": 4032
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00027267106347897776,
+      "loss": 4.4575,
+      "step": 4033
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00027246496290189617,
+      "loss": 4.1608,
+      "step": 4034
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002722588623248145,
+      "loss": 4.4019,
+      "step": 4035
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00027205276174773293,
+      "loss": 4.0353,
+      "step": 4036
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002718466611706513,
+      "loss": 4.1233,
+      "step": 4037
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002716405605935697,
+      "loss": 4.2052,
+      "step": 4038
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00027143446001648804,
+      "loss": 3.9711,
+      "step": 4039
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00027122835943940645,
+      "loss": 4.9555,
+      "step": 4040
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00027102225886232485,
+      "loss": 4.0783,
+      "step": 4041
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002708161582852432,
+      "loss": 4.4239,
+      "step": 4042
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002706100577081616,
+      "loss": 4.5244,
+      "step": 4043
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00027040395713107997,
+      "loss": 4.2096,
+      "step": 4044
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00027019785655399837,
+      "loss": 4.2045,
+      "step": 4045
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002699917559769167,
+      "loss": 4.3807,
+      "step": 4046
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026978565539983513,
+      "loss": 4.0367,
+      "step": 4047
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026957955482275354,
+      "loss": 4.5302,
+      "step": 4048
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002693734542456719,
+      "loss": 4.0669,
+      "step": 4049
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002691673536685903,
+      "loss": 4.6156,
+      "step": 4050
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026896125309150865,
+      "loss": 4.1794,
+      "step": 4051
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026875515251442706,
+      "loss": 4.1865,
+      "step": 4052
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002685490519373454,
+      "loss": 4.6166,
+      "step": 4053
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002683429513602638,
+      "loss": 4.1507,
+      "step": 4054
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026813685078318217,
+      "loss": 4.0992,
+      "step": 4055
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0002679307502061006,
+      "loss": 4.1403,
+      "step": 4056
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000267724649629019,
+      "loss": 4.4724,
+      "step": 4057
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00026751854905193733,
+      "loss": 4.2966,
+      "step": 4058
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00026731244847485574,
+      "loss": 4.1615,
+      "step": 4059
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002671063478977741,
+      "loss": 4.2871,
+      "step": 4060
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002669002473206925,
+      "loss": 4.4085,
+      "step": 4061
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00026669414674361085,
+      "loss": 3.8434,
+      "step": 4062
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00026648804616652926,
+      "loss": 4.2188,
+      "step": 4063
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00026628194558944767,
+      "loss": 4.4806,
+      "step": 4064
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000266075845012366,
+      "loss": 3.9369,
+      "step": 4065
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002658697444352844,
+      "loss": 3.9501,
+      "step": 4066
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002656636438582028,
+      "loss": 3.9133,
+      "step": 4067
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002654575432811212,
+      "loss": 4.35,
+      "step": 4068
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00026525144270403954,
+      "loss": 4.5714,
+      "step": 4069
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00026504534212695794,
+      "loss": 4.43,
+      "step": 4070
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002648392415498763,
+      "loss": 4.3218,
+      "step": 4071
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002646331409727947,
+      "loss": 4.279,
+      "step": 4072
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00026442704039571316,
+      "loss": 4.1303,
+      "step": 4073
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002642209398186315,
+      "loss": 4.6159,
+      "step": 4074
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002640148392415499,
+      "loss": 4.4837,
+      "step": 4075
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002638087386644683,
+      "loss": 4.8308,
+      "step": 4076
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002636026380873867,
+      "loss": 4.4826,
+      "step": 4077
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026339653751030503,
+      "loss": 4.2648,
+      "step": 4078
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026319043693322344,
+      "loss": 4.4638,
+      "step": 4079
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026298433635614185,
+      "loss": 4.1065,
+      "step": 4080
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002627782357790602,
+      "loss": 4.0786,
+      "step": 4081
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002625721352019786,
+      "loss": 4.1024,
+      "step": 4082
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026236603462489696,
+      "loss": 4.3444,
+      "step": 4083
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026215993404781537,
+      "loss": 4.2388,
+      "step": 4084
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002619538334707337,
+      "loss": 4.2415,
+      "step": 4085
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002617477328936521,
+      "loss": 4.4211,
+      "step": 4086
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002615416323165705,
+      "loss": 4.3097,
+      "step": 4087
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002613355317394889,
+      "loss": 4.3474,
+      "step": 4088
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002611294311624073,
+      "loss": 4.3197,
+      "step": 4089
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026092333058532564,
+      "loss": 4.2487,
+      "step": 4090
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026071723000824405,
+      "loss": 4.2947,
+      "step": 4091
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002605111294311624,
+      "loss": 4.1389,
+      "step": 4092
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002603050288540808,
+      "loss": 4.3099,
+      "step": 4093
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00026009892827699916,
+      "loss": 4.4808,
+      "step": 4094
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025989282769991757,
+      "loss": 4.1923,
+      "step": 4095
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.000259686727122836,
+      "loss": 4.3825,
+      "step": 4096
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025948062654575433,
+      "loss": 4.48,
+      "step": 4097
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025927452596867273,
+      "loss": 4.3302,
+      "step": 4098
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002590684253915911,
+      "loss": 4.2465,
+      "step": 4099
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002588623248145095,
+      "loss": 4.4367,
+      "step": 4100
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025865622423742785,
+      "loss": 4.2989,
+      "step": 4101
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025845012366034625,
+      "loss": 3.9492,
+      "step": 4102
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002582440230832646,
+      "loss": 4.0314,
+      "step": 4103
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.000258037922506183,
+      "loss": 4.0453,
+      "step": 4104
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002578318219291014,
+      "loss": 4.0945,
+      "step": 4105
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025762572135201977,
+      "loss": 3.9574,
+      "step": 4106
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002574196207749382,
+      "loss": 4.1493,
+      "step": 4107
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025721352019785653,
+      "loss": 4.015,
+      "step": 4108
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00025700741962077494,
+      "loss": 4.3152,
+      "step": 4109
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002568013190436933,
+      "loss": 4.2219,
+      "step": 4110
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002565952184666117,
+      "loss": 4.2844,
+      "step": 4111
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002563891178895301,
+      "loss": 3.6476,
+      "step": 4112
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00025618301731244845,
+      "loss": 3.7629,
+      "step": 4113
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00025597691673536686,
+      "loss": 4.8554,
+      "step": 4114
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002557708161582852,
+      "loss": 3.7332,
+      "step": 4115
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002555647155812036,
+      "loss": 3.875,
+      "step": 4116
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.000255358615004122,
+      "loss": 3.9219,
+      "step": 4117
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002551525144270404,
+      "loss": 3.7086,
+      "step": 4118
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00025494641384995873,
+      "loss": 4.3901,
+      "step": 4119
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002547403132728772,
+      "loss": 4.6572,
+      "step": 4120
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002545342126957956,
+      "loss": 4.5239,
+      "step": 4121
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00025432811211871395,
+      "loss": 4.2452,
+      "step": 4122
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00025412201154163236,
+      "loss": 4.1438,
+      "step": 4123
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002539159109645507,
+      "loss": 4.3398,
+      "step": 4124
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002537098103874691,
+      "loss": 4.2189,
+      "step": 4125
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00025350370981038747,
+      "loss": 4.3408,
+      "step": 4126
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002532976092333059,
+      "loss": 4.1881,
+      "step": 4127
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0002530915086562243,
+      "loss": 4.4415,
+      "step": 4128
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00025288540807914264,
+      "loss": 4.3039,
+      "step": 4129
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00025267930750206104,
+      "loss": 4.4467,
+      "step": 4130
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002524732069249794,
+      "loss": 4.602,
+      "step": 4131
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002522671063478978,
+      "loss": 4.3033,
+      "step": 4132
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00025206100577081616,
+      "loss": 4.8008,
+      "step": 4133
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00025185490519373456,
+      "loss": 4.3715,
+      "step": 4134
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002516488046166529,
+      "loss": 4.1163,
+      "step": 4135
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002514427040395713,
+      "loss": 4.2486,
+      "step": 4136
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00025123660346248973,
+      "loss": 4.3878,
+      "step": 4137
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002510305028854081,
+      "loss": 4.0991,
+      "step": 4138
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002508244023083265,
+      "loss": 4.0861,
+      "step": 4139
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00025061830173124484,
+      "loss": 4.259,
+      "step": 4140
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00025041220115416325,
+      "loss": 4.2381,
+      "step": 4141
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002502061005770816,
+      "loss": 4.1878,
+      "step": 4142
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00025,
+      "loss": 4.0906,
+      "step": 4143
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002497938994229184,
+      "loss": 4.5765,
+      "step": 4144
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00024958779884583676,
+      "loss": 4.1133,
+      "step": 4145
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00024938169826875517,
+      "loss": 4.3547,
+      "step": 4146
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002491755976916736,
+      "loss": 4.2872,
+      "step": 4147
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024896949711459193,
+      "loss": 4.3639,
+      "step": 4148
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024876339653751034,
+      "loss": 4.256,
+      "step": 4149
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002485572959604287,
+      "loss": 4.0568,
+      "step": 4150
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002483511953833471,
+      "loss": 4.1395,
+      "step": 4151
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024814509480626545,
+      "loss": 4.0236,
+      "step": 4152
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024793899422918386,
+      "loss": 4.1359,
+      "step": 4153
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002477328936521022,
+      "loss": 4.2059,
+      "step": 4154
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002475267930750206,
+      "loss": 3.8551,
+      "step": 4155
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.000247320692497939,
+      "loss": 4.3433,
+      "step": 4156
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002471145919208574,
+      "loss": 3.9827,
+      "step": 4157
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002469084913437758,
+      "loss": 4.0275,
+      "step": 4158
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024670239076669413,
+      "loss": 3.9997,
+      "step": 4159
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024649629018961254,
+      "loss": 4.1793,
+      "step": 4160
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002462901896125309,
+      "loss": 4.3352,
+      "step": 4161
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002460840890354493,
+      "loss": 4.1006,
+      "step": 4162
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0002458779884583677,
+      "loss": 4.2637,
+      "step": 4163
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024567188788128606,
+      "loss": 4.2218,
+      "step": 4164
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00024546578730420446,
+      "loss": 4.0474,
+      "step": 4165
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002452596867271228,
+      "loss": 3.9252,
+      "step": 4166
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002450535861500412,
+      "loss": 3.7889,
+      "step": 4167
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024484748557295963,
+      "loss": 3.4926,
+      "step": 4168
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.000244641384995878,
+      "loss": 4.411,
+      "step": 4169
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002444352844187964,
+      "loss": 4.1841,
+      "step": 4170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002442291838417148,
+      "loss": 4.7075,
+      "step": 4171
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024402308326463315,
+      "loss": 4.3475,
+      "step": 4172
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024381698268755153,
+      "loss": 4.2657,
+      "step": 4173
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002436108821104699,
+      "loss": 4.2714,
+      "step": 4174
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0002434047815333883,
+      "loss": 4.4266,
+      "step": 4175
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024319868095630667,
+      "loss": 4.2939,
+      "step": 4176
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024299258037922505,
+      "loss": 4.6089,
+      "step": 4177
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024278647980214343,
+      "loss": 4.7966,
+      "step": 4178
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024258037922506186,
+      "loss": 3.9867,
+      "step": 4179
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024237427864798024,
+      "loss": 4.554,
+      "step": 4180
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024216817807089862,
+      "loss": 4.7748,
+      "step": 4181
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.000241962077493817,
+      "loss": 4.3421,
+      "step": 4182
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00024175597691673538,
+      "loss": 4.4205,
+      "step": 4183
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00024154987633965376,
+      "loss": 4.4796,
+      "step": 4184
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00024134377576257214,
+      "loss": 3.9604,
+      "step": 4185
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00024113767518549052,
+      "loss": 4.3006,
+      "step": 4186
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00024093157460840892,
+      "loss": 4.247,
+      "step": 4187
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002407254740313273,
+      "loss": 4.05,
+      "step": 4188
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00024051937345424568,
+      "loss": 4.4522,
+      "step": 4189
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00024031327287716406,
+      "loss": 4.3291,
+      "step": 4190
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00024010717230008244,
+      "loss": 4.3547,
+      "step": 4191
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00023990107172300082,
+      "loss": 4.0893,
+      "step": 4192
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002396949711459192,
+      "loss": 4.2819,
+      "step": 4193
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00023948887056883758,
+      "loss": 4.3224,
+      "step": 4194
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000239282769991756,
+      "loss": 4.2546,
+      "step": 4195
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00023907666941467437,
+      "loss": 4.1279,
+      "step": 4196
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00023887056883759275,
+      "loss": 4.0951,
+      "step": 4197
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00023866446826051113,
+      "loss": 4.5426,
+      "step": 4198
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0002384583676834295,
+      "loss": 4.1469,
+      "step": 4199
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00023825226710634789,
+      "loss": 4.0152,
+      "step": 4200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00023804616652926627,
+      "loss": 4.0652,
+      "step": 4201
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023784006595218467,
+      "loss": 4.4396,
+      "step": 4202
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023763396537510308,
+      "loss": 4.2069,
+      "step": 4203
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023742786479802146,
+      "loss": 4.2655,
+      "step": 4204
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023722176422093984,
+      "loss": 4.2513,
+      "step": 4205
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023701566364385822,
+      "loss": 4.2226,
+      "step": 4206
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002368095630667766,
+      "loss": 4.3578,
+      "step": 4207
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023660346248969498,
+      "loss": 4.1541,
+      "step": 4208
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023639736191261336,
+      "loss": 4.4234,
+      "step": 4209
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023619126133553174,
+      "loss": 4.3,
+      "step": 4210
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023598516075845014,
+      "loss": 4.416,
+      "step": 4211
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023577906018136852,
+      "loss": 3.796,
+      "step": 4212
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002355729596042869,
+      "loss": 3.9267,
+      "step": 4213
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023536685902720528,
+      "loss": 3.6895,
+      "step": 4214
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023516075845012366,
+      "loss": 4.3174,
+      "step": 4215
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023495465787304204,
+      "loss": 4.0764,
+      "step": 4216
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023474855729596042,
+      "loss": 3.9326,
+      "step": 4217
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002345424567188788,
+      "loss": 4.0156,
+      "step": 4218
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002343363561417972,
+      "loss": 4.1861,
+      "step": 4219
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023413025556471559,
+      "loss": 4.4891,
+      "step": 4220
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023392415498763397,
+      "loss": 4.6352,
+      "step": 4221
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023371805441055234,
+      "loss": 4.4473,
+      "step": 4222
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023351195383347072,
+      "loss": 4.0394,
+      "step": 4223
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002333058532563891,
+      "loss": 4.1792,
+      "step": 4224
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002330997526793075,
+      "loss": 4.0698,
+      "step": 4225
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002328936521022259,
+      "loss": 4.2196,
+      "step": 4226
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002326875515251443,
+      "loss": 4.2794,
+      "step": 4227
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023248145094806268,
+      "loss": 4.5937,
+      "step": 4228
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023227535037098106,
+      "loss": 4.205,
+      "step": 4229
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023206924979389944,
+      "loss": 4.2345,
+      "step": 4230
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023186314921681782,
+      "loss": 4.517,
+      "step": 4231
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002316570486397362,
+      "loss": 4.2438,
+      "step": 4232
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023145094806265457,
+      "loss": 4.0339,
+      "step": 4233
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023124484748557295,
+      "loss": 4.2525,
+      "step": 4234
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023103874690849136,
+      "loss": 4.161,
+      "step": 4235
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023083264633140974,
+      "loss": 4.2535,
+      "step": 4236
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00023062654575432812,
+      "loss": 4.2505,
+      "step": 4237
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002304204451772465,
+      "loss": 4.238,
+      "step": 4238
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00023021434460016488,
+      "loss": 4.2825,
+      "step": 4239
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00023000824402308326,
+      "loss": 4.2342,
+      "step": 4240
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00022980214344600164,
+      "loss": 4.2478,
+      "step": 4241
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00022959604286892002,
+      "loss": 4.4171,
+      "step": 4242
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00022938994229183842,
+      "loss": 4.164,
+      "step": 4243
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002291838417147568,
+      "loss": 4.5729,
+      "step": 4244
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00022897774113767518,
+      "loss": 4.082,
+      "step": 4245
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00022877164056059356,
+      "loss": 4.0886,
+      "step": 4246
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00022856553998351194,
+      "loss": 4.6268,
+      "step": 4247
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00022835943940643032,
+      "loss": 4.2723,
+      "step": 4248
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00022815333882934873,
+      "loss": 4.2524,
+      "step": 4249
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002279472382522671,
+      "loss": 3.9722,
+      "step": 4250
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00022774113767518552,
+      "loss": 4.1651,
+      "step": 4251
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0002275350370981039,
+      "loss": 3.9598,
+      "step": 4252
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00022732893652102227,
+      "loss": 4.0955,
+      "step": 4253
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00022712283594394065,
+      "loss": 4.1968,
+      "step": 4254
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00022691673536685903,
+      "loss": 4.1806,
+      "step": 4255
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002267106347897774,
+      "loss": 3.8498,
+      "step": 4256
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002265045342126958,
+      "loss": 3.9873,
+      "step": 4257
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002262984336356142,
+      "loss": 4.2745,
+      "step": 4258
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00022609233305853258,
+      "loss": 4.2401,
+      "step": 4259
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00022588623248145096,
+      "loss": 3.8629,
+      "step": 4260
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00022568013190436934,
+      "loss": 3.9743,
+      "step": 4261
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00022547403132728772,
+      "loss": 4.1664,
+      "step": 4262
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002252679307502061,
+      "loss": 3.8293,
+      "step": 4263
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00022506183017312448,
+      "loss": 3.7872,
+      "step": 4264
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00022485572959604286,
+      "loss": 3.94,
+      "step": 4265
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00022464962901896126,
+      "loss": 3.5326,
+      "step": 4266
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00022444352844187964,
+      "loss": 3.7027,
+      "step": 4267
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00022423742786479802,
+      "loss": 3.9701,
+      "step": 4268
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0002240313272877164,
+      "loss": 4.4647,
+      "step": 4269
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00022382522671063478,
+      "loss": 4.4842,
+      "step": 4270
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00022361912613355316,
+      "loss": 4.3476,
+      "step": 4271
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00022341302555647157,
+      "loss": 4.0815,
+      "step": 4272
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00022320692497938995,
+      "loss": 4.5096,
+      "step": 4273
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00022300082440230835,
+      "loss": 4.2815,
+      "step": 4274
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00022279472382522673,
+      "loss": 4.126,
+      "step": 4275
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00022258862324814511,
+      "loss": 4.6363,
+      "step": 4276
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002223825226710635,
+      "loss": 4.2135,
+      "step": 4277
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00022217642209398187,
+      "loss": 4.188,
+      "step": 4278
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00022197032151690025,
+      "loss": 4.3018,
+      "step": 4279
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00022176422093981863,
+      "loss": 4.4145,
+      "step": 4280
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.000221558120362737,
+      "loss": 4.4261,
+      "step": 4281
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00022135201978565542,
+      "loss": 4.1601,
+      "step": 4282
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002211459192085738,
+      "loss": 4.3668,
+      "step": 4283
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00022093981863149218,
+      "loss": 4.4448,
+      "step": 4284
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00022073371805441056,
+      "loss": 4.5648,
+      "step": 4285
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00022052761747732894,
+      "loss": 4.1012,
+      "step": 4286
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00022032151690024732,
+      "loss": 4.1254,
+      "step": 4287
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0002201154163231657,
+      "loss": 4.1805,
+      "step": 4288
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00021990931574608408,
+      "loss": 4.2902,
+      "step": 4289
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00021970321516900248,
+      "loss": 4.2214,
+      "step": 4290
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00021949711459192086,
+      "loss": 4.3349,
+      "step": 4291
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00021929101401483924,
+      "loss": 4.0498,
+      "step": 4292
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00021908491343775762,
+      "loss": 4.2793,
+      "step": 4293
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.000218878812860676,
+      "loss": 4.4776,
+      "step": 4294
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002186727122835944,
+      "loss": 4.1622,
+      "step": 4295
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002184666117065128,
+      "loss": 4.1901,
+      "step": 4296
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00021826051112943117,
+      "loss": 4.0475,
+      "step": 4297
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00021805441055234957,
+      "loss": 4.5885,
+      "step": 4298
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00021784830997526795,
+      "loss": 4.0039,
+      "step": 4299
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00021764220939818633,
+      "loss": 4.2109,
+      "step": 4300
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002174361088211047,
+      "loss": 4.2225,
+      "step": 4301
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002172300082440231,
+      "loss": 4.0679,
+      "step": 4302
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00021702390766694147,
+      "loss": 4.6083,
+      "step": 4303
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00021681780708985985,
+      "loss": 4.049,
+      "step": 4304
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00021661170651277823,
+      "loss": 4.2275,
+      "step": 4305
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00021640560593569664,
+      "loss": 4.1763,
+      "step": 4306
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00021619950535861502,
+      "loss": 4.0712,
+      "step": 4307
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002159934047815334,
+      "loss": 4.1209,
+      "step": 4308
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00021578730420445178,
+      "loss": 4.1123,
+      "step": 4309
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00021558120362737016,
+      "loss": 4.3178,
+      "step": 4310
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00021537510305028853,
+      "loss": 4.6948,
+      "step": 4311
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00021516900247320691,
+      "loss": 4.35,
+      "step": 4312
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002149629018961253,
+      "loss": 4.1688,
+      "step": 4313
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002147568013190437,
+      "loss": 4.0774,
+      "step": 4314
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00021455070074196208,
+      "loss": 4.5166,
+      "step": 4315
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00021434460016488046,
+      "loss": 4.1338,
+      "step": 4316
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00021413849958779884,
+      "loss": 3.7864,
+      "step": 4317
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00021393239901071725,
+      "loss": 4.3624,
+      "step": 4318
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00021372629843363563,
+      "loss": 4.342,
+      "step": 4319
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.000213520197856554,
+      "loss": 4.7782,
+      "step": 4320
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00021331409727947238,
+      "loss": 4.1658,
+      "step": 4321
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002131079967023908,
+      "loss": 4.4519,
+      "step": 4322
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00021290189612530917,
+      "loss": 4.2087,
+      "step": 4323
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00021269579554822755,
+      "loss": 4.2655,
+      "step": 4324
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00021248969497114593,
+      "loss": 4.2188,
+      "step": 4325
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002122835943940643,
+      "loss": 4.6617,
+      "step": 4326
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002120774938169827,
+      "loss": 4.5083,
+      "step": 4327
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00021187139323990107,
+      "loss": 4.0348,
+      "step": 4328
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00021166529266281945,
+      "loss": 4.193,
+      "step": 4329
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00021145919208573786,
+      "loss": 4.1286,
+      "step": 4330
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00021125309150865623,
+      "loss": 4.3174,
+      "step": 4331
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00021104699093157461,
+      "loss": 4.4024,
+      "step": 4332
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.000210840890354493,
+      "loss": 4.2824,
+      "step": 4333
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00021063478977741137,
+      "loss": 4.3501,
+      "step": 4334
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00021042868920032975,
+      "loss": 4.2824,
+      "step": 4335
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00021022258862324813,
+      "loss": 4.4705,
+      "step": 4336
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002100164880461665,
+      "loss": 4.204,
+      "step": 4337
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00020981038746908492,
+      "loss": 4.3012,
+      "step": 4338
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002096042868920033,
+      "loss": 4.6911,
+      "step": 4339
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00020939818631492168,
+      "loss": 4.2352,
+      "step": 4340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00020919208573784006,
+      "loss": 4.2209,
+      "step": 4341
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00020898598516075846,
+      "loss": 4.2214,
+      "step": 4342
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00020877988458367684,
+      "loss": 4.2201,
+      "step": 4343
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00020857378400659522,
+      "loss": 4.5116,
+      "step": 4344
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002083676834295136,
+      "loss": 4.2347,
+      "step": 4345
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.000208161582852432,
+      "loss": 4.039,
+      "step": 4346
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002079554822753504,
+      "loss": 4.2854,
+      "step": 4347
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00020774938169826877,
+      "loss": 4.294,
+      "step": 4348
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00020754328112118715,
+      "loss": 4.302,
+      "step": 4349
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00020733718054410553,
+      "loss": 4.0458,
+      "step": 4350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002071310799670239,
+      "loss": 4.0327,
+      "step": 4351
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002069249793899423,
+      "loss": 3.9848,
+      "step": 4352
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00020671887881286067,
+      "loss": 4.1688,
+      "step": 4353
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00020651277823577907,
+      "loss": 4.114,
+      "step": 4354
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00020630667765869745,
+      "loss": 4.1521,
+      "step": 4355
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00020610057708161583,
+      "loss": 4.1181,
+      "step": 4356
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002058944765045342,
+      "loss": 4.3291,
+      "step": 4357
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002056883759274526,
+      "loss": 4.2059,
+      "step": 4358
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00020548227535037097,
+      "loss": 4.158,
+      "step": 4359
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00020527617477328935,
+      "loss": 4.4022,
+      "step": 4360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00020507007419620773,
+      "loss": 4.1354,
+      "step": 4361
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00020486397361912614,
+      "loss": 4.3799,
+      "step": 4362
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00020465787304204452,
+      "loss": 4.0402,
+      "step": 4363
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002044517724649629,
+      "loss": 4.1853,
+      "step": 4364
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002042456718878813,
+      "loss": 4.0858,
+      "step": 4365
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00020403957131079968,
+      "loss": 3.6809,
+      "step": 4366
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00020383347073371806,
+      "loss": 3.851,
+      "step": 4367
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00020362737015663644,
+      "loss": 3.4608,
+      "step": 4368
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00020342126957955485,
+      "loss": 4.5582,
+      "step": 4369
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00020321516900247323,
+      "loss": 4.2923,
+      "step": 4370
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002030090684253916,
+      "loss": 4.3933,
+      "step": 4371
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00020280296784831,
+      "loss": 4.2827,
+      "step": 4372
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00020259686727122837,
+      "loss": 4.1224,
+      "step": 4373
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00020239076669414675,
+      "loss": 4.2496,
+      "step": 4374
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00020218466611706513,
+      "loss": 4.3954,
+      "step": 4375
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002019785655399835,
+      "loss": 4.2258,
+      "step": 4376
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002017724649629019,
+      "loss": 4.1596,
+      "step": 4377
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002015663643858203,
+      "loss": 4.4959,
+      "step": 4378
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00020136026380873867,
+      "loss": 4.4638,
+      "step": 4379
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00020115416323165705,
+      "loss": 4.275,
+      "step": 4380
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00020094806265457543,
+      "loss": 4.0906,
+      "step": 4381
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002007419620774938,
+      "loss": 4.2308,
+      "step": 4382
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002005358615004122,
+      "loss": 4.2008,
+      "step": 4383
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00020032976092333057,
+      "loss": 4.1375,
+      "step": 4384
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00020012366034624898,
+      "loss": 4.2948,
+      "step": 4385
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00019991755976916736,
+      "loss": 3.9726,
+      "step": 4386
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00019971145919208574,
+      "loss": 4.2705,
+      "step": 4387
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00019950535861500414,
+      "loss": 4.1187,
+      "step": 4388
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00019929925803792252,
+      "loss": 4.2397,
+      "step": 4389
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001990931574608409,
+      "loss": 4.3624,
+      "step": 4390
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00019888705688375928,
+      "loss": 3.9631,
+      "step": 4391
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00019868095630667766,
+      "loss": 4.167,
+      "step": 4392
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00019847485572959607,
+      "loss": 4.1483,
+      "step": 4393
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00019826875515251445,
+      "loss": 4.0655,
+      "step": 4394
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00019806265457543283,
+      "loss": 4.3051,
+      "step": 4395
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001978565539983512,
+      "loss": 3.9631,
+      "step": 4396
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00019765045342126959,
+      "loss": 4.6478,
+      "step": 4397
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00019744435284418797,
+      "loss": 4.3562,
+      "step": 4398
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00019723825226710634,
+      "loss": 4.3334,
+      "step": 4399
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00019703215169002472,
+      "loss": 4.0571,
+      "step": 4400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00019682605111294313,
+      "loss": 4.1191,
+      "step": 4401
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001966199505358615,
+      "loss": 4.123,
+      "step": 4402
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001964138499587799,
+      "loss": 3.9015,
+      "step": 4403
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00019620774938169827,
+      "loss": 4.3845,
+      "step": 4404
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00019600164880461665,
+      "loss": 4.2594,
+      "step": 4405
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00019579554822753503,
+      "loss": 4.0342,
+      "step": 4406
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001955894476504534,
+      "loss": 4.291,
+      "step": 4407
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001953833470733718,
+      "loss": 3.8477,
+      "step": 4408
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001951772464962902,
+      "loss": 4.141,
+      "step": 4409
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00019497114591920857,
+      "loss": 4.1299,
+      "step": 4410
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00019476504534212695,
+      "loss": 4.2989,
+      "step": 4411
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00019455894476504536,
+      "loss": 3.9486,
+      "step": 4412
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00019435284418796374,
+      "loss": 4.0931,
+      "step": 4413
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00019414674361088212,
+      "loss": 4.2363,
+      "step": 4414
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001939406430338005,
+      "loss": 4.0309,
+      "step": 4415
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019373454245671888,
+      "loss": 3.7799,
+      "step": 4416
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019352844187963729,
+      "loss": 4.4142,
+      "step": 4417
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019332234130255567,
+      "loss": 3.8759,
+      "step": 4418
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019311624072547404,
+      "loss": 4.3088,
+      "step": 4419
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019291014014839242,
+      "loss": 4.2382,
+      "step": 4420
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001927040395713108,
+      "loss": 4.3945,
+      "step": 4421
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019249793899422918,
+      "loss": 4.1754,
+      "step": 4422
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019229183841714756,
+      "loss": 4.259,
+      "step": 4423
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019208573784006594,
+      "loss": 4.0778,
+      "step": 4424
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019187963726298435,
+      "loss": 4.291,
+      "step": 4425
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019167353668590273,
+      "loss": 4.3008,
+      "step": 4426
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001914674361088211,
+      "loss": 4.3513,
+      "step": 4427
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001912613355317395,
+      "loss": 4.3779,
+      "step": 4428
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019105523495465787,
+      "loss": 4.4821,
+      "step": 4429
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019084913437757625,
+      "loss": 4.2129,
+      "step": 4430
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00019064303380049463,
+      "loss": 4.1125,
+      "step": 4431
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000190436933223413,
+      "loss": 4.1096,
+      "step": 4432
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001902308326463314,
+      "loss": 4.2545,
+      "step": 4433
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001900247320692498,
+      "loss": 4.2241,
+      "step": 4434
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001898186314921682,
+      "loss": 4.1882,
+      "step": 4435
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00018961253091508658,
+      "loss": 4.2101,
+      "step": 4436
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00018940643033800496,
+      "loss": 4.0668,
+      "step": 4437
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00018920032976092334,
+      "loss": 4.3124,
+      "step": 4438
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00018899422918384172,
+      "loss": 4.2241,
+      "step": 4439
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001887881286067601,
+      "loss": 4.2782,
+      "step": 4440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001885820280296785,
+      "loss": 4.2137,
+      "step": 4441
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00018837592745259688,
+      "loss": 4.1216,
+      "step": 4442
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00018816982687551526,
+      "loss": 4.382,
+      "step": 4443
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00018796372629843364,
+      "loss": 4.0812,
+      "step": 4444
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00018775762572135202,
+      "loss": 4.3123,
+      "step": 4445
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001875515251442704,
+      "loss": 4.2419,
+      "step": 4446
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00018734542456718878,
+      "loss": 4.0422,
+      "step": 4447
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00018713932399010716,
+      "loss": 4.3368,
+      "step": 4448
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00018693322341302557,
+      "loss": 4.5692,
+      "step": 4449
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00018672712283594395,
+      "loss": 4.1233,
+      "step": 4450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00018652102225886233,
+      "loss": 4.1314,
+      "step": 4451
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001863149216817807,
+      "loss": 4.0347,
+      "step": 4452
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018610882110469909,
+      "loss": 4.1971,
+      "step": 4453
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018590272052761747,
+      "loss": 4.4273,
+      "step": 4454
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018569661995053585,
+      "loss": 4.3287,
+      "step": 4455
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018549051937345423,
+      "loss": 4.4279,
+      "step": 4456
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018528441879637263,
+      "loss": 4.2038,
+      "step": 4457
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018507831821929104,
+      "loss": 4.0663,
+      "step": 4458
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018487221764220942,
+      "loss": 4.2542,
+      "step": 4459
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001846661170651278,
+      "loss": 4.4417,
+      "step": 4460
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018446001648804618,
+      "loss": 4.0599,
+      "step": 4461
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018425391591096456,
+      "loss": 3.6868,
+      "step": 4462
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018404781533388294,
+      "loss": 3.9921,
+      "step": 4463
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018384171475680132,
+      "loss": 3.9783,
+      "step": 4464
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018363561417971972,
+      "loss": 4.1348,
+      "step": 4465
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001834295136026381,
+      "loss": 4.2697,
+      "step": 4466
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018322341302555648,
+      "loss": 4.0306,
+      "step": 4467
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00018301731244847486,
+      "loss": 3.6299,
+      "step": 4468
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00018281121187139324,
+      "loss": 4.2602,
+      "step": 4469
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00018260511129431162,
+      "loss": 4.4423,
+      "step": 4470
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00018239901071723,
+      "loss": 4.2326,
+      "step": 4471
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00018219291014014838,
+      "loss": 4.8139,
+      "step": 4472
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001819868095630668,
+      "loss": 4.016,
+      "step": 4473
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00018178070898598517,
+      "loss": 4.241,
+      "step": 4474
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00018157460840890355,
+      "loss": 4.0527,
+      "step": 4475
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00018136850783182193,
+      "loss": 4.1664,
+      "step": 4476
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001811624072547403,
+      "loss": 4.0906,
+      "step": 4477
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00018095630667765868,
+      "loss": 4.3045,
+      "step": 4478
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00018075020610057706,
+      "loss": 4.4604,
+      "step": 4479
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00018054410552349547,
+      "loss": 4.1096,
+      "step": 4480
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00018033800494641385,
+      "loss": 4.3087,
+      "step": 4481
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00018013190436933226,
+      "loss": 4.2578,
+      "step": 4482
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00017992580379225064,
+      "loss": 3.966,
+      "step": 4483
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00017971970321516902,
+      "loss": 4.0461,
+      "step": 4484
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001795136026380874,
+      "loss": 4.0757,
+      "step": 4485
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00017930750206100578,
+      "loss": 4.1429,
+      "step": 4486
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017910140148392416,
+      "loss": 4.0525,
+      "step": 4487
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017889530090684256,
+      "loss": 4.2442,
+      "step": 4488
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017868920032976094,
+      "loss": 4.135,
+      "step": 4489
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017848309975267932,
+      "loss": 4.3974,
+      "step": 4490
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001782769991755977,
+      "loss": 4.4497,
+      "step": 4491
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017807089859851608,
+      "loss": 4.3231,
+      "step": 4492
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017786479802143446,
+      "loss": 4.3893,
+      "step": 4493
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017765869744435284,
+      "loss": 4.3605,
+      "step": 4494
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017745259686727122,
+      "loss": 4.0287,
+      "step": 4495
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017724649629018963,
+      "loss": 4.3291,
+      "step": 4496
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.000177040395713108,
+      "loss": 4.1944,
+      "step": 4497
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017683429513602638,
+      "loss": 4.1548,
+      "step": 4498
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017662819455894476,
+      "loss": 3.9312,
+      "step": 4499
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017642209398186314,
+      "loss": 4.1189,
+      "step": 4500
+    },
+    {
+      "epoch": 2.52,
+      "eval_loss": 5.279923915863037,
+      "eval_runtime": 1245.8116,
+      "eval_samples_per_second": 2.121,
+      "eval_steps_per_second": 0.266,
+      "eval_wer": 1.9154303847679492,
+      "step": 4500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017621599340478152,
+      "loss": 4.1845,
+      "step": 4501
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001760098928276999,
+      "loss": 4.3047,
+      "step": 4502
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00017580379225061828,
+      "loss": 4.3513,
+      "step": 4503
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001755976916735367,
+      "loss": 4.1046,
+      "step": 4504
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001753915910964551,
+      "loss": 4.403,
+      "step": 4505
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017518549051937348,
+      "loss": 3.8846,
+      "step": 4506
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017497938994229186,
+      "loss": 3.9162,
+      "step": 4507
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017477328936521023,
+      "loss": 4.3908,
+      "step": 4508
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017456718878812861,
+      "loss": 4.2078,
+      "step": 4509
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.000174361088211047,
+      "loss": 4.2066,
+      "step": 4510
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017415498763396537,
+      "loss": 4.1656,
+      "step": 4511
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017394888705688378,
+      "loss": 4.1172,
+      "step": 4512
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017374278647980216,
+      "loss": 4.1159,
+      "step": 4513
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017353668590272054,
+      "loss": 4.5064,
+      "step": 4514
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017333058532563892,
+      "loss": 3.988,
+      "step": 4515
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001731244847485573,
+      "loss": 3.8681,
+      "step": 4516
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017291838417147568,
+      "loss": 3.3975,
+      "step": 4517
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017271228359439406,
+      "loss": 3.2584,
+      "step": 4518
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017250618301731244,
+      "loss": 4.8464,
+      "step": 4519
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017230008244023084,
+      "loss": 4.3045,
+      "step": 4520
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00017209398186314922,
+      "loss": 4.6907,
+      "step": 4521
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001718878812860676,
+      "loss": 4.2102,
+      "step": 4522
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00017168178070898598,
+      "loss": 4.1382,
+      "step": 4523
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00017147568013190436,
+      "loss": 4.5249,
+      "step": 4524
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00017126957955482274,
+      "loss": 4.3217,
+      "step": 4525
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00017106347897774112,
+      "loss": 4.3966,
+      "step": 4526
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001708573784006595,
+      "loss": 4.3098,
+      "step": 4527
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00017065127782357793,
+      "loss": 4.4944,
+      "step": 4528
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00017044517724649631,
+      "loss": 4.2748,
+      "step": 4529
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001702390766694147,
+      "loss": 4.3319,
+      "step": 4530
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00017003297609233307,
+      "loss": 4.3716,
+      "step": 4531
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00016982687551525145,
+      "loss": 4.3954,
+      "step": 4532
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00016962077493816983,
+      "loss": 4.0397,
+      "step": 4533
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001694146743610882,
+      "loss": 4.2336,
+      "step": 4534
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001692085737840066,
+      "loss": 4.0153,
+      "step": 4535
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.000169002473206925,
+      "loss": 4.3044,
+      "step": 4536
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00016879637262984338,
+      "loss": 4.396,
+      "step": 4537
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00016859027205276176,
+      "loss": 4.2732,
+      "step": 4538
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00016838417147568014,
+      "loss": 4.1545,
+      "step": 4539
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00016817807089859852,
+      "loss": 4.2093,
+      "step": 4540
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001679719703215169,
+      "loss": 4.2182,
+      "step": 4541
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016776586974443528,
+      "loss": 4.1137,
+      "step": 4542
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016755976916735366,
+      "loss": 4.2818,
+      "step": 4543
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016735366859027206,
+      "loss": 4.1284,
+      "step": 4544
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016714756801319044,
+      "loss": 4.1199,
+      "step": 4545
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016694146743610882,
+      "loss": 3.8086,
+      "step": 4546
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001667353668590272,
+      "loss": 4.3677,
+      "step": 4547
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016652926628194558,
+      "loss": 4.5244,
+      "step": 4548
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016632316570486396,
+      "loss": 3.9887,
+      "step": 4549
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016611706512778234,
+      "loss": 4.2151,
+      "step": 4550
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016591096455070072,
+      "loss": 4.3198,
+      "step": 4551
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016570486397361915,
+      "loss": 4.2235,
+      "step": 4552
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016549876339653753,
+      "loss": 4.0355,
+      "step": 4553
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001652926628194559,
+      "loss": 4.3388,
+      "step": 4554
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001650865622423743,
+      "loss": 3.9741,
+      "step": 4555
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016488046166529267,
+      "loss": 4.0154,
+      "step": 4556
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016467436108821105,
+      "loss": 4.1872,
+      "step": 4557
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00016446826051112943,
+      "loss": 4.081,
+      "step": 4558
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001642621599340478,
+      "loss": 3.8857,
+      "step": 4559
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016405605935696622,
+      "loss": 3.935,
+      "step": 4560
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001638499587798846,
+      "loss": 4.1418,
+      "step": 4561
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016364385820280298,
+      "loss": 3.9302,
+      "step": 4562
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016343775762572136,
+      "loss": 3.893,
+      "step": 4563
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016323165704863974,
+      "loss": 4.103,
+      "step": 4564
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016302555647155812,
+      "loss": 4.0528,
+      "step": 4565
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001628194558944765,
+      "loss": 3.8769,
+      "step": 4566
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016261335531739487,
+      "loss": 4.1166,
+      "step": 4567
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016240725474031328,
+      "loss": 4.1454,
+      "step": 4568
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016220115416323166,
+      "loss": 4.1994,
+      "step": 4569
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016199505358615004,
+      "loss": 4.1412,
+      "step": 4570
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016178895300906842,
+      "loss": 4.259,
+      "step": 4571
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001615828524319868,
+      "loss": 4.1854,
+      "step": 4572
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016137675185490518,
+      "loss": 4.2057,
+      "step": 4573
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016117065127782356,
+      "loss": 4.2208,
+      "step": 4574
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016096455070074197,
+      "loss": 4.1216,
+      "step": 4575
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00016075845012366037,
+      "loss": 4.4522,
+      "step": 4576
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00016055234954657875,
+      "loss": 4.2137,
+      "step": 4577
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00016034624896949713,
+      "loss": 4.2191,
+      "step": 4578
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001601401483924155,
+      "loss": 4.3187,
+      "step": 4579
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001599340478153339,
+      "loss": 3.9219,
+      "step": 4580
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00015972794723825227,
+      "loss": 4.4043,
+      "step": 4581
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00015952184666117065,
+      "loss": 4.0578,
+      "step": 4582
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00015931574608408903,
+      "loss": 4.0249,
+      "step": 4583
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00015910964550700744,
+      "loss": 4.1554,
+      "step": 4584
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00015890354492992582,
+      "loss": 4.4685,
+      "step": 4585
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001586974443528442,
+      "loss": 4.0377,
+      "step": 4586
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00015849134377576257,
+      "loss": 4.7574,
+      "step": 4587
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00015828524319868095,
+      "loss": 3.9672,
+      "step": 4588
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00015807914262159933,
+      "loss": 4.3722,
+      "step": 4589
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001578730420445177,
+      "loss": 4.1472,
+      "step": 4590
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00015766694146743612,
+      "loss": 4.2962,
+      "step": 4591
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001574608408903545,
+      "loss": 4.3331,
+      "step": 4592
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00015725474031327288,
+      "loss": 4.1165,
+      "step": 4593
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00015704863973619126,
+      "loss": 4.1792,
+      "step": 4594
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00015684253915910964,
+      "loss": 4.1127,
+      "step": 4595
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00015663643858202802,
+      "loss": 4.0441,
+      "step": 4596
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001564303380049464,
+      "loss": 3.9547,
+      "step": 4597
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00015622423742786478,
+      "loss": 4.3655,
+      "step": 4598
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001560181368507832,
+      "loss": 4.1221,
+      "step": 4599
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001558120362737016,
+      "loss": 4.0603,
+      "step": 4600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00015560593569661997,
+      "loss": 4.2452,
+      "step": 4601
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00015539983511953835,
+      "loss": 3.9755,
+      "step": 4602
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00015519373454245673,
+      "loss": 4.224,
+      "step": 4603
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001549876339653751,
+      "loss": 3.8009,
+      "step": 4604
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001547815333882935,
+      "loss": 4.4653,
+      "step": 4605
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00015457543281121187,
+      "loss": 4.8057,
+      "step": 4606
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00015436933223413027,
+      "loss": 4.1858,
+      "step": 4607
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00015416323165704865,
+      "loss": 3.9762,
+      "step": 4608
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00015395713107996703,
+      "loss": 4.1375,
+      "step": 4609
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001537510305028854,
+      "loss": 4.1704,
+      "step": 4610
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001535449299258038,
+      "loss": 3.9839,
+      "step": 4611
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015333882934872217,
+      "loss": 3.8314,
+      "step": 4612
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015313272877164055,
+      "loss": 4.1102,
+      "step": 4613
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015292662819455893,
+      "loss": 3.5205,
+      "step": 4614
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015272052761747734,
+      "loss": 3.6952,
+      "step": 4615
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015251442704039572,
+      "loss": 3.7667,
+      "step": 4616
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001523083264633141,
+      "loss": 3.8582,
+      "step": 4617
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015210222588623248,
+      "loss": 3.2266,
+      "step": 4618
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015189612530915086,
+      "loss": 4.7796,
+      "step": 4619
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015169002473206924,
+      "loss": 4.4603,
+      "step": 4620
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015148392415498762,
+      "loss": 4.1939,
+      "step": 4621
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015127782357790602,
+      "loss": 4.3807,
+      "step": 4622
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015107172300082443,
+      "loss": 4.2668,
+      "step": 4623
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001508656224237428,
+      "loss": 4.437,
+      "step": 4624
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001506595218466612,
+      "loss": 4.3236,
+      "step": 4625
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015045342126957957,
+      "loss": 4.3601,
+      "step": 4626
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015024732069249795,
+      "loss": 4.2584,
+      "step": 4627
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00015004122011541633,
+      "loss": 4.3483,
+      "step": 4628
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001498351195383347,
+      "loss": 4.745,
+      "step": 4629
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014962901896125309,
+      "loss": 4.3689,
+      "step": 4630
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001494229183841715,
+      "loss": 4.1145,
+      "step": 4631
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014921681780708987,
+      "loss": 4.4887,
+      "step": 4632
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014901071723000825,
+      "loss": 4.1417,
+      "step": 4633
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014880461665292663,
+      "loss": 4.2087,
+      "step": 4634
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000148598516075845,
+      "loss": 4.022,
+      "step": 4635
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001483924154987634,
+      "loss": 4.1217,
+      "step": 4636
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014818631492168177,
+      "loss": 4.2019,
+      "step": 4637
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014798021434460015,
+      "loss": 4.1869,
+      "step": 4638
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014777411376751856,
+      "loss": 4.2881,
+      "step": 4639
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014756801319043694,
+      "loss": 4.0914,
+      "step": 4640
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014736191261335532,
+      "loss": 3.9886,
+      "step": 4641
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001471558120362737,
+      "loss": 4.4184,
+      "step": 4642
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014694971145919208,
+      "loss": 4.4915,
+      "step": 4643
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014674361088211045,
+      "loss": 4.0062,
+      "step": 4644
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014653751030502886,
+      "loss": 4.0311,
+      "step": 4645
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014633140972794724,
+      "loss": 4.5158,
+      "step": 4646
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014612530915086565,
+      "loss": 4.154,
+      "step": 4647
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014591920857378403,
+      "loss": 4.0143,
+      "step": 4648
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001457131079967024,
+      "loss": 4.2906,
+      "step": 4649
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001455070074196208,
+      "loss": 3.8838,
+      "step": 4650
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014530090684253917,
+      "loss": 4.2958,
+      "step": 4651
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014509480626545755,
+      "loss": 3.8014,
+      "step": 4652
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014488870568837593,
+      "loss": 4.2503,
+      "step": 4653
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001446826051112943,
+      "loss": 4.3844,
+      "step": 4654
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001444765045342127,
+      "loss": 4.0721,
+      "step": 4655
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001442704039571311,
+      "loss": 4.356,
+      "step": 4656
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014406430338004947,
+      "loss": 4.1843,
+      "step": 4657
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014385820280296785,
+      "loss": 4.1055,
+      "step": 4658
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014365210222588623,
+      "loss": 3.9479,
+      "step": 4659
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001434460016488046,
+      "loss": 3.8631,
+      "step": 4660
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.000143239901071723,
+      "loss": 4.1113,
+      "step": 4661
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014303380049464137,
+      "loss": 3.9033,
+      "step": 4662
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014282769991755978,
+      "loss": 3.7628,
+      "step": 4663
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014262159934047815,
+      "loss": 4.1743,
+      "step": 4664
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014241549876339653,
+      "loss": 3.8201,
+      "step": 4665
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014220939818631491,
+      "loss": 3.8355,
+      "step": 4666
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001420032976092333,
+      "loss": 3.9487,
+      "step": 4667
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001417971970321517,
+      "loss": 3.5505,
+      "step": 4668
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014159109645507008,
+      "loss": 4.3465,
+      "step": 4669
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014138499587798846,
+      "loss": 4.3526,
+      "step": 4670
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014117889530090687,
+      "loss": 4.5241,
+      "step": 4671
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014097279472382525,
+      "loss": 4.2153,
+      "step": 4672
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014076669414674363,
+      "loss": 3.8713,
+      "step": 4673
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000140560593569662,
+      "loss": 4.3129,
+      "step": 4674
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014035449299258038,
+      "loss": 4.3615,
+      "step": 4675
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014014839241549876,
+      "loss": 4.3204,
+      "step": 4676
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00013994229183841714,
+      "loss": 4.0873,
+      "step": 4677
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00013973619126133552,
+      "loss": 4.0874,
+      "step": 4678
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00013953009068425393,
+      "loss": 4.3027,
+      "step": 4679
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001393239901071723,
+      "loss": 4.2762,
+      "step": 4680
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001391178895300907,
+      "loss": 4.1922,
+      "step": 4681
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00013891178895300907,
+      "loss": 3.9317,
+      "step": 4682
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00013870568837592745,
+      "loss": 4.3559,
+      "step": 4683
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013849958779884583,
+      "loss": 4.0347,
+      "step": 4684
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001382934872217642,
+      "loss": 4.3221,
+      "step": 4685
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001380873866446826,
+      "loss": 4.1614,
+      "step": 4686
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000137881286067601,
+      "loss": 4.1853,
+      "step": 4687
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013767518549051937,
+      "loss": 3.9605,
+      "step": 4688
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013746908491343775,
+      "loss": 4.231,
+      "step": 4689
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013726298433635613,
+      "loss": 3.9214,
+      "step": 4690
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001370568837592745,
+      "loss": 4.4512,
+      "step": 4691
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013685078318219292,
+      "loss": 4.3055,
+      "step": 4692
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001366446826051113,
+      "loss": 4.1906,
+      "step": 4693
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013643858202802968,
+      "loss": 4.253,
+      "step": 4694
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013623248145094808,
+      "loss": 4.0658,
+      "step": 4695
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013602638087386646,
+      "loss": 4.241,
+      "step": 4696
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013582028029678484,
+      "loss": 4.0585,
+      "step": 4697
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013561417971970322,
+      "loss": 4.2588,
+      "step": 4698
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001354080791426216,
+      "loss": 4.3119,
+      "step": 4699
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013520197856553998,
+      "loss": 3.987,
+      "step": 4700
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013499587798845836,
+      "loss": 4.1778,
+      "step": 4701
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013478977741137677,
+      "loss": 4.0324,
+      "step": 4702
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013458367683429515,
+      "loss": 4.0837,
+      "step": 4703
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013437757625721353,
+      "loss": 4.0803,
+      "step": 4704
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001341714756801319,
+      "loss": 4.4023,
+      "step": 4705
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001339653751030503,
+      "loss": 4.2164,
+      "step": 4706
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013375927452596867,
+      "loss": 4.2004,
+      "step": 4707
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013355317394888705,
+      "loss": 4.3101,
+      "step": 4708
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013334707337180543,
+      "loss": 3.9467,
+      "step": 4709
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013314097279472383,
+      "loss": 4.2424,
+      "step": 4710
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001329348722176422,
+      "loss": 3.7424,
+      "step": 4711
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001327287716405606,
+      "loss": 3.9002,
+      "step": 4712
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013252267106347897,
+      "loss": 3.5189,
+      "step": 4713
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013231657048639735,
+      "loss": 3.981,
+      "step": 4714
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013211046990931576,
+      "loss": 3.5417,
+      "step": 4715
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013190436933223414,
+      "loss": 4.0744,
+      "step": 4716
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013169826875515252,
+      "loss": 5.0636,
+      "step": 4717
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013149216817807092,
+      "loss": 3.9172,
+      "step": 4718
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001312860676009893,
+      "loss": 4.6544,
+      "step": 4719
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013107996702390768,
+      "loss": 4.2603,
+      "step": 4720
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013087386644682606,
+      "loss": 4.1938,
+      "step": 4721
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013066776586974444,
+      "loss": 4.1439,
+      "step": 4722
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013046166529266282,
+      "loss": 4.1226,
+      "step": 4723
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001302555647155812,
+      "loss": 4.5914,
+      "step": 4724
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013004946413849958,
+      "loss": 4.3083,
+      "step": 4725
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.000129843363561418,
+      "loss": 4.2255,
+      "step": 4726
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012963726298433637,
+      "loss": 4.1675,
+      "step": 4727
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012943116240725475,
+      "loss": 4.009,
+      "step": 4728
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012922506183017313,
+      "loss": 4.481,
+      "step": 4729
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001290189612530915,
+      "loss": 4.4863,
+      "step": 4730
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012881286067600989,
+      "loss": 4.0326,
+      "step": 4731
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012860676009892826,
+      "loss": 4.4124,
+      "step": 4732
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012840065952184664,
+      "loss": 4.1262,
+      "step": 4733
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012819455894476505,
+      "loss": 4.1352,
+      "step": 4734
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012798845836768343,
+      "loss": 4.4019,
+      "step": 4735
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001277823577906018,
+      "loss": 4.3238,
+      "step": 4736
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001275762572135202,
+      "loss": 3.9994,
+      "step": 4737
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001273701566364386,
+      "loss": 3.9704,
+      "step": 4738
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012716405605935698,
+      "loss": 4.1264,
+      "step": 4739
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012695795548227536,
+      "loss": 3.9828,
+      "step": 4740
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012675185490519374,
+      "loss": 4.2245,
+      "step": 4741
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012654575432811214,
+      "loss": 3.8472,
+      "step": 4742
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012633965375103052,
+      "loss": 4.1452,
+      "step": 4743
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001261335531739489,
+      "loss": 4.206,
+      "step": 4744
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012592745259686728,
+      "loss": 4.0286,
+      "step": 4745
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012572135201978566,
+      "loss": 4.2402,
+      "step": 4746
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012551525144270404,
+      "loss": 4.2526,
+      "step": 4747
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012530915086562242,
+      "loss": 4.0118,
+      "step": 4748
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001251030502885408,
+      "loss": 4.1121,
+      "step": 4749
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001248969497114592,
+      "loss": 4.3763,
+      "step": 4750
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012469084913437759,
+      "loss": 4.2174,
+      "step": 4751
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012448474855729597,
+      "loss": 3.9423,
+      "step": 4752
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012427864798021434,
+      "loss": 4.1951,
+      "step": 4753
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012407254740313272,
+      "loss": 3.9706,
+      "step": 4754
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001238664468260511,
+      "loss": 4.1729,
+      "step": 4755
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001236603462489695,
+      "loss": 4.2926,
+      "step": 4756
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001234542456718879,
+      "loss": 4.1748,
+      "step": 4757
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012324814509480627,
+      "loss": 3.8936,
+      "step": 4758
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012304204451772465,
+      "loss": 4.039,
+      "step": 4759
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012283594394064303,
+      "loss": 4.4066,
+      "step": 4760
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001226298433635614,
+      "loss": 3.9811,
+      "step": 4761
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012242374278647982,
+      "loss": 4.0079,
+      "step": 4762
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001222176422093982,
+      "loss": 3.7391,
+      "step": 4763
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012201154163231657,
+      "loss": 3.9014,
+      "step": 4764
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012180544105523495,
+      "loss": 4.0588,
+      "step": 4765
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012159934047815333,
+      "loss": 4.0663,
+      "step": 4766
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012139323990107171,
+      "loss": 4.1089,
+      "step": 4767
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012118713932399012,
+      "loss": 3.0831,
+      "step": 4768
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001209810387469085,
+      "loss": 4.4483,
+      "step": 4769
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012077493816982688,
+      "loss": 4.3244,
+      "step": 4770
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012056883759274526,
+      "loss": 4.2301,
+      "step": 4771
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012036273701566365,
+      "loss": 4.2391,
+      "step": 4772
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00012015663643858203,
+      "loss": 4.0639,
+      "step": 4773
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011995053586150041,
+      "loss": 4.0781,
+      "step": 4774
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011974443528441879,
+      "loss": 4.2461,
+      "step": 4775
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011953833470733718,
+      "loss": 4.0588,
+      "step": 4776
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011933223413025556,
+      "loss": 4.4475,
+      "step": 4777
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011912613355317394,
+      "loss": 3.9267,
+      "step": 4778
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011892003297609234,
+      "loss": 4.1295,
+      "step": 4779
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011871393239901073,
+      "loss": 4.0926,
+      "step": 4780
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011850783182192911,
+      "loss": 4.0312,
+      "step": 4781
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011830173124484749,
+      "loss": 4.2004,
+      "step": 4782
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011809563066776587,
+      "loss": 3.7264,
+      "step": 4783
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011788953009068426,
+      "loss": 3.8747,
+      "step": 4784
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011768342951360264,
+      "loss": 4.3976,
+      "step": 4785
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011747732893652102,
+      "loss": 4.0522,
+      "step": 4786
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001172712283594394,
+      "loss": 4.3864,
+      "step": 4787
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011706512778235779,
+      "loss": 4.2434,
+      "step": 4788
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011685902720527617,
+      "loss": 4.022,
+      "step": 4789
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00011665292662819455,
+      "loss": 4.2119,
+      "step": 4790
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00011644682605111295,
+      "loss": 4.0395,
+      "step": 4791
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00011624072547403134,
+      "loss": 4.3781,
+      "step": 4792
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00011603462489694972,
+      "loss": 3.9749,
+      "step": 4793
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001158285243198681,
+      "loss": 4.0094,
+      "step": 4794
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00011562242374278648,
+      "loss": 4.0509,
+      "step": 4795
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00011541632316570487,
+      "loss": 3.9379,
+      "step": 4796
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00011521022258862325,
+      "loss": 4.1694,
+      "step": 4797
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00011500412201154163,
+      "loss": 4.3636,
+      "step": 4798
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00011479802143446001,
+      "loss": 4.1538,
+      "step": 4799
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001145919208573784,
+      "loss": 4.2361,
+      "step": 4800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00011438582028029678,
+      "loss": 4.0,
+      "step": 4801
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00011417971970321516,
+      "loss": 4.7772,
+      "step": 4802
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00011397361912613355,
+      "loss": 3.9351,
+      "step": 4803
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00011376751854905195,
+      "loss": 3.8935,
+      "step": 4804
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00011356141797197033,
+      "loss": 4.4024,
+      "step": 4805
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001133553173948887,
+      "loss": 3.9298,
+      "step": 4806
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001131492168178071,
+      "loss": 3.9579,
+      "step": 4807
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011294311624072548,
+      "loss": 3.8717,
+      "step": 4808
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011273701566364386,
+      "loss": 4.1512,
+      "step": 4809
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011253091508656224,
+      "loss": 4.1998,
+      "step": 4810
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011232481450948063,
+      "loss": 3.9852,
+      "step": 4811
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011211871393239901,
+      "loss": 4.14,
+      "step": 4812
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011191261335531739,
+      "loss": 4.1613,
+      "step": 4813
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011170651277823578,
+      "loss": 3.9355,
+      "step": 4814
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011150041220115418,
+      "loss": 3.6314,
+      "step": 4815
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011129431162407256,
+      "loss": 3.632,
+      "step": 4816
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011108821104699094,
+      "loss": 4.2708,
+      "step": 4817
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011088211046990932,
+      "loss": 3.7348,
+      "step": 4818
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011067600989282771,
+      "loss": 4.4972,
+      "step": 4819
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011046990931574609,
+      "loss": 4.4637,
+      "step": 4820
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011026380873866447,
+      "loss": 4.1775,
+      "step": 4821
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00011005770816158285,
+      "loss": 4.1112,
+      "step": 4822
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00010985160758450124,
+      "loss": 4.0556,
+      "step": 4823
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00010964550700741962,
+      "loss": 4.2926,
+      "step": 4824
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000109439406430338,
+      "loss": 4.2509,
+      "step": 4825
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001092333058532564,
+      "loss": 4.0573,
+      "step": 4826
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010902720527617479,
+      "loss": 4.5802,
+      "step": 4827
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010882110469909317,
+      "loss": 3.999,
+      "step": 4828
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010861500412201155,
+      "loss": 4.0847,
+      "step": 4829
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010840890354492993,
+      "loss": 4.3032,
+      "step": 4830
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010820280296784832,
+      "loss": 3.9546,
+      "step": 4831
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001079967023907667,
+      "loss": 4.1293,
+      "step": 4832
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010779060181368508,
+      "loss": 4.4338,
+      "step": 4833
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010758450123660346,
+      "loss": 4.1544,
+      "step": 4834
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010737840065952185,
+      "loss": 4.1444,
+      "step": 4835
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010717230008244023,
+      "loss": 4.1497,
+      "step": 4836
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010696619950535862,
+      "loss": 4.2753,
+      "step": 4837
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000106760098928277,
+      "loss": 4.0076,
+      "step": 4838
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001065539983511954,
+      "loss": 4.0726,
+      "step": 4839
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010634789777411378,
+      "loss": 4.0948,
+      "step": 4840
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010614179719703215,
+      "loss": 4.3616,
+      "step": 4841
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010593569661995053,
+      "loss": 4.2832,
+      "step": 4842
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00010572959604286893,
+      "loss": 3.9787,
+      "step": 4843
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010552349546578731,
+      "loss": 4.0474,
+      "step": 4844
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010531739488870569,
+      "loss": 4.0602,
+      "step": 4845
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010511129431162407,
+      "loss": 4.3332,
+      "step": 4846
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010490519373454246,
+      "loss": 3.6797,
+      "step": 4847
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010469909315746084,
+      "loss": 4.0424,
+      "step": 4848
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010449299258037923,
+      "loss": 4.3258,
+      "step": 4849
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010428689200329761,
+      "loss": 4.2831,
+      "step": 4850
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.000104080791426216,
+      "loss": 3.9246,
+      "step": 4851
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010387469084913438,
+      "loss": 4.2392,
+      "step": 4852
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010366859027205276,
+      "loss": 3.8803,
+      "step": 4853
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010346248969497114,
+      "loss": 4.5585,
+      "step": 4854
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010325638911788954,
+      "loss": 4.2185,
+      "step": 4855
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010305028854080792,
+      "loss": 4.3447,
+      "step": 4856
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001028441879637263,
+      "loss": 4.1393,
+      "step": 4857
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010263808738664468,
+      "loss": 4.068,
+      "step": 4858
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010243198680956307,
+      "loss": 3.9345,
+      "step": 4859
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010222588623248145,
+      "loss": 3.9692,
+      "step": 4860
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00010201978565539984,
+      "loss": 3.8957,
+      "step": 4861
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00010181368507831822,
+      "loss": 4.3485,
+      "step": 4862
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00010160758450123661,
+      "loss": 3.8649,
+      "step": 4863
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000101401483924155,
+      "loss": 4.1485,
+      "step": 4864
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00010119538334707337,
+      "loss": 3.5227,
+      "step": 4865
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00010098928276999175,
+      "loss": 4.1999,
+      "step": 4866
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00010078318219291015,
+      "loss": 4.2115,
+      "step": 4867
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00010057708161582853,
+      "loss": 3.6373,
+      "step": 4868
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001003709810387469,
+      "loss": 4.4264,
+      "step": 4869
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00010016488046166528,
+      "loss": 4.2233,
+      "step": 4870
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.995877988458368e-05,
+      "loss": 4.3151,
+      "step": 4871
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.975267930750207e-05,
+      "loss": 4.2595,
+      "step": 4872
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.954657873042045e-05,
+      "loss": 4.2636,
+      "step": 4873
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.934047815333883e-05,
+      "loss": 4.0441,
+      "step": 4874
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.913437757625722e-05,
+      "loss": 4.3673,
+      "step": 4875
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.89282769991756e-05,
+      "loss": 4.0972,
+      "step": 4876
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.872217642209398e-05,
+      "loss": 4.2057,
+      "step": 4877
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.851607584501236e-05,
+      "loss": 4.048,
+      "step": 4878
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.830997526793076e-05,
+      "loss": 4.0784,
+      "step": 4879
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.810387469084913e-05,
+      "loss": 4.3603,
+      "step": 4880
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.789777411376751e-05,
+      "loss": 4.1404,
+      "step": 4881
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.76916735366859e-05,
+      "loss": 4.3495,
+      "step": 4882
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.748557295960429e-05,
+      "loss": 4.195,
+      "step": 4883
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.727947238252268e-05,
+      "loss": 4.0602,
+      "step": 4884
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.707337180544106e-05,
+      "loss": 4.53,
+      "step": 4885
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.686727122835944e-05,
+      "loss": 4.02,
+      "step": 4886
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.666117065127783e-05,
+      "loss": 4.2864,
+      "step": 4887
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.645507007419621e-05,
+      "loss": 3.9559,
+      "step": 4888
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.624896949711459e-05,
+      "loss": 4.5806,
+      "step": 4889
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.604286892003297e-05,
+      "loss": 4.0812,
+      "step": 4890
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.583676834295136e-05,
+      "loss": 4.101,
+      "step": 4891
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.563066776586974e-05,
+      "loss": 4.2905,
+      "step": 4892
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.542456718878812e-05,
+      "loss": 4.2536,
+      "step": 4893
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.52184666117065e-05,
+      "loss": 3.9487,
+      "step": 4894
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.50123660346249e-05,
+      "loss": 4.0374,
+      "step": 4895
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.480626545754329e-05,
+      "loss": 4.0602,
+      "step": 4896
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.460016488046167e-05,
+      "loss": 4.0316,
+      "step": 4897
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.439406430338005e-05,
+      "loss": 4.2802,
+      "step": 4898
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.418796372629844e-05,
+      "loss": 4.0329,
+      "step": 4899
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.398186314921682e-05,
+      "loss": 4.1439,
+      "step": 4900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.37757625721352e-05,
+      "loss": 4.0784,
+      "step": 4901
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.356966199505358e-05,
+      "loss": 4.4323,
+      "step": 4902
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.336356141797197e-05,
+      "loss": 4.0,
+      "step": 4903
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.315746084089035e-05,
+      "loss": 3.9314,
+      "step": 4904
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.295136026380873e-05,
+      "loss": 4.0797,
+      "step": 4905
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.274525968672711e-05,
+      "loss": 3.8776,
+      "step": 4906
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.253915910964552e-05,
+      "loss": 3.9849,
+      "step": 4907
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.23330585325639e-05,
+      "loss": 4.0567,
+      "step": 4908
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.212695795548228e-05,
+      "loss": 3.8793,
+      "step": 4909
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.192085737840066e-05,
+      "loss": 4.2522,
+      "step": 4910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.171475680131905e-05,
+      "loss": 4.1134,
+      "step": 4911
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.150865622423743e-05,
+      "loss": 4.1321,
+      "step": 4912
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.130255564715581e-05,
+      "loss": 3.7761,
+      "step": 4913
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.109645507007419e-05,
+      "loss": 4.0619,
+      "step": 4914
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.089035449299258e-05,
+      "loss": 3.6459,
+      "step": 4915
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.068425391591096e-05,
+      "loss": 3.625,
+      "step": 4916
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.047815333882934e-05,
+      "loss": 4.0972,
+      "step": 4917
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.027205276174774e-05,
+      "loss": 3.8633,
+      "step": 4918
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.006595218466613e-05,
+      "loss": 4.2141,
+      "step": 4919
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.985985160758451e-05,
+      "loss": 4.1884,
+      "step": 4920
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.965375103050289e-05,
+      "loss": 4.4533,
+      "step": 4921
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.944765045342128e-05,
+      "loss": 4.5156,
+      "step": 4922
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.924154987633966e-05,
+      "loss": 4.1187,
+      "step": 4923
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.903544929925804e-05,
+      "loss": 4.3381,
+      "step": 4924
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.882934872217642e-05,
+      "loss": 4.1952,
+      "step": 4925
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.862324814509481e-05,
+      "loss": 4.173,
+      "step": 4926
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.841714756801319e-05,
+      "loss": 4.159,
+      "step": 4927
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.821104699093157e-05,
+      "loss": 4.4774,
+      "step": 4928
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.800494641384995e-05,
+      "loss": 4.0497,
+      "step": 4929
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.779884583676834e-05,
+      "loss": 4.0904,
+      "step": 4930
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.759274525968674e-05,
+      "loss": 4.3767,
+      "step": 4931
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.738664468260512e-05,
+      "loss": 4.6125,
+      "step": 4932
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.71805441055235e-05,
+      "loss": 4.3799,
+      "step": 4933
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.697444352844189e-05,
+      "loss": 4.2351,
+      "step": 4934
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.676834295136027e-05,
+      "loss": 4.1793,
+      "step": 4935
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.656224237427865e-05,
+      "loss": 4.5071,
+      "step": 4936
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.635614179719703e-05,
+      "loss": 4.1743,
+      "step": 4937
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.615004122011542e-05,
+      "loss": 4.2663,
+      "step": 4938
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.59439406430338e-05,
+      "loss": 4.192,
+      "step": 4939
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.573784006595218e-05,
+      "loss": 4.1668,
+      "step": 4940
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.553173948887056e-05,
+      "loss": 4.1609,
+      "step": 4941
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.532563891178897e-05,
+      "loss": 4.0529,
+      "step": 4942
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.511953833470735e-05,
+      "loss": 4.0009,
+      "step": 4943
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.491343775762573e-05,
+      "loss": 4.3228,
+      "step": 4944
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.47073371805441e-05,
+      "loss": 4.0608,
+      "step": 4945
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.45012366034625e-05,
+      "loss": 4.1112,
+      "step": 4946
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.429513602638088e-05,
+      "loss": 4.012,
+      "step": 4947
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.408903544929926e-05,
+      "loss": 4.7125,
+      "step": 4948
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.388293487221764e-05,
+      "loss": 4.2506,
+      "step": 4949
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.367683429513603e-05,
+      "loss": 4.104,
+      "step": 4950
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.347073371805441e-05,
+      "loss": 4.2745,
+      "step": 4951
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.326463314097279e-05,
+      "loss": 4.1237,
+      "step": 4952
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.305853256389117e-05,
+      "loss": 4.1495,
+      "step": 4953
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.285243198680958e-05,
+      "loss": 3.8845,
+      "step": 4954
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.264633140972796e-05,
+      "loss": 4.099,
+      "step": 4955
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.244023083264634e-05,
+      "loss": 3.9419,
+      "step": 4956
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.223413025556472e-05,
+      "loss": 4.0972,
+      "step": 4957
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.202802967848311e-05,
+      "loss": 4.1548,
+      "step": 4958
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.182192910140149e-05,
+      "loss": 4.0662,
+      "step": 4959
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.161582852431987e-05,
+      "loss": 4.3084,
+      "step": 4960
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.140972794723825e-05,
+      "loss": 3.8023,
+      "step": 4961
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.120362737015664e-05,
+      "loss": 3.8764,
+      "step": 4962
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.099752679307502e-05,
+      "loss": 3.883,
+      "step": 4963
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.07914262159934e-05,
+      "loss": 4.3458,
+      "step": 4964
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.058532563891178e-05,
+      "loss": 3.5909,
+      "step": 4965
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.037922506183019e-05,
+      "loss": 3.7131,
+      "step": 4966
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.017312448474857e-05,
+      "loss": 3.9905,
+      "step": 4967
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.996702390766695e-05,
+      "loss": 3.6367,
+      "step": 4968
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.976092333058532e-05,
+      "loss": 4.1407,
+      "step": 4969
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.955482275350372e-05,
+      "loss": 4.1235,
+      "step": 4970
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.93487221764221e-05,
+      "loss": 4.4319,
+      "step": 4971
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.914262159934048e-05,
+      "loss": 4.1425,
+      "step": 4972
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.893652102225886e-05,
+      "loss": 4.1644,
+      "step": 4973
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.873042044517725e-05,
+      "loss": 3.9834,
+      "step": 4974
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.852431986809563e-05,
+      "loss": 4.0235,
+      "step": 4975
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.831821929101401e-05,
+      "loss": 4.2807,
+      "step": 4976
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.811211871393239e-05,
+      "loss": 3.9996,
+      "step": 4977
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.79060181368508e-05,
+      "loss": 4.4148,
+      "step": 4978
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.769991755976917e-05,
+      "loss": 4.2735,
+      "step": 4979
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.749381698268755e-05,
+      "loss": 4.1779,
+      "step": 4980
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.728771640560593e-05,
+      "loss": 4.2022,
+      "step": 4981
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.708161582852433e-05,
+      "loss": 3.8674,
+      "step": 4982
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.68755152514427e-05,
+      "loss": 4.3302,
+      "step": 4983
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.666941467436109e-05,
+      "loss": 4.0902,
+      "step": 4984
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.646331409727947e-05,
+      "loss": 4.1535,
+      "step": 4985
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.625721352019786e-05,
+      "loss": 4.1151,
+      "step": 4986
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.605111294311624e-05,
+      "loss": 3.9085,
+      "step": 4987
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.584501236603462e-05,
+      "loss": 4.1388,
+      "step": 4988
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.563891178895301e-05,
+      "loss": 4.0986,
+      "step": 4989
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.54328112118714e-05,
+      "loss": 4.3576,
+      "step": 4990
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.522671063478978e-05,
+      "loss": 4.1769,
+      "step": 4991
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.502061005770816e-05,
+      "loss": 3.9775,
+      "step": 4992
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.481450948062654e-05,
+      "loss": 4.1916,
+      "step": 4993
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.460840890354494e-05,
+      "loss": 4.1726,
+      "step": 4994
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.440230832646332e-05,
+      "loss": 4.343,
+      "step": 4995
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.41962077493817e-05,
+      "loss": 3.8344,
+      "step": 4996
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.399010717230008e-05,
+      "loss": 4.1213,
+      "step": 4997
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.378400659521847e-05,
+      "loss": 4.0898,
+      "step": 4998
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.357790601813685e-05,
+      "loss": 4.4431,
+      "step": 4999
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.337180544105523e-05,
+      "loss": 3.9921,
+      "step": 5000
+    },
+    {
+      "epoch": 2.8,
+      "eval_loss": 5.245349884033203,
+      "eval_runtime": 1249.1915,
+      "eval_samples_per_second": 2.115,
+      "eval_steps_per_second": 0.265,
+      "eval_wer": 1.906981356604522,
+      "step": 5000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.316570486397362e-05,
+      "loss": 4.1309,
+      "step": 5001
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.295960428689201e-05,
+      "loss": 4.2133,
+      "step": 5002
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.27535037098104e-05,
+      "loss": 4.024,
+      "step": 5003
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.254740313272877e-05,
+      "loss": 4.0509,
+      "step": 5004
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.234130255564715e-05,
+      "loss": 4.3849,
+      "step": 5005
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.213520197856555e-05,
+      "loss": 4.2124,
+      "step": 5006
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.192910140148393e-05,
+      "loss": 4.0046,
+      "step": 5007
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.17230008244023e-05,
+      "loss": 4.1701,
+      "step": 5008
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.151690024732068e-05,
+      "loss": 3.8308,
+      "step": 5009
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.131079967023908e-05,
+      "loss": 4.0012,
+      "step": 5010
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.110469909315746e-05,
+      "loss": 4.5083,
+      "step": 5011
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.089859851607585e-05,
+      "loss": 4.1657,
+      "step": 5012
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.069249793899423e-05,
+      "loss": 3.8427,
+      "step": 5013
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.048639736191262e-05,
+      "loss": 3.9107,
+      "step": 5014
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.0280296784831e-05,
+      "loss": 3.8355,
+      "step": 5015
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.007419620774938e-05,
+      "loss": 4.0505,
+      "step": 5016
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.986809563066776e-05,
+      "loss": 3.9924,
+      "step": 5017
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.966199505358615e-05,
+      "loss": 3.6692,
+      "step": 5018
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.945589447650453e-05,
+      "loss": 4.1603,
+      "step": 5019
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.924979389942291e-05,
+      "loss": 4.1836,
+      "step": 5020
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.90436933223413e-05,
+      "loss": 4.3312,
+      "step": 5021
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.883759274525969e-05,
+      "loss": 4.3707,
+      "step": 5022
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.863149216817807e-05,
+      "loss": 4.4177,
+      "step": 5023
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.842539159109646e-05,
+      "loss": 4.0264,
+      "step": 5024
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.821929101401484e-05,
+      "loss": 4.0478,
+      "step": 5025
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.801319043693323e-05,
+      "loss": 4.3421,
+      "step": 5026
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.780708985985161e-05,
+      "loss": 3.9696,
+      "step": 5027
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.760098928276999e-05,
+      "loss": 4.2373,
+      "step": 5028
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.739488870568838e-05,
+      "loss": 4.3438,
+      "step": 5029
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.718878812860676e-05,
+      "loss": 4.0777,
+      "step": 5030
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.698268755152514e-05,
+      "loss": 4.4004,
+      "step": 5031
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.677658697444352e-05,
+      "loss": 3.9597,
+      "step": 5032
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.657048639736192e-05,
+      "loss": 4.1742,
+      "step": 5033
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.63643858202803e-05,
+      "loss": 3.7932,
+      "step": 5034
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.615828524319868e-05,
+      "loss": 4.0445,
+      "step": 5035
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.595218466611707e-05,
+      "loss": 4.0306,
+      "step": 5036
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.574608408903546e-05,
+      "loss": 3.9919,
+      "step": 5037
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.553998351195384e-05,
+      "loss": 4.0238,
+      "step": 5038
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.533388293487222e-05,
+      "loss": 4.1692,
+      "step": 5039
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.51277823577906e-05,
+      "loss": 4.2674,
+      "step": 5040
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.4921681780709e-05,
+      "loss": 3.9967,
+      "step": 5041
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.471558120362737e-05,
+      "loss": 4.118,
+      "step": 5042
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.450948062654575e-05,
+      "loss": 4.5215,
+      "step": 5043
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.430338004946413e-05,
+      "loss": 4.3447,
+      "step": 5044
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.409727947238253e-05,
+      "loss": 3.8581,
+      "step": 5045
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.38911788953009e-05,
+      "loss": 4.2783,
+      "step": 5046
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.36850783182193e-05,
+      "loss": 4.1119,
+      "step": 5047
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.347897774113768e-05,
+      "loss": 4.0754,
+      "step": 5048
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.327287716405607e-05,
+      "loss": 4.0416,
+      "step": 5049
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.306677658697445e-05,
+      "loss": 4.3116,
+      "step": 5050
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.286067600989283e-05,
+      "loss": 4.0366,
+      "step": 5051
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.265457543281121e-05,
+      "loss": 4.0488,
+      "step": 5052
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.24484748557296e-05,
+      "loss": 4.0539,
+      "step": 5053
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.224237427864798e-05,
+      "loss": 3.8755,
+      "step": 5054
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.203627370156636e-05,
+      "loss": 4.2029,
+      "step": 5055
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.183017312448476e-05,
+      "loss": 3.8834,
+      "step": 5056
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.162407254740313e-05,
+      "loss": 4.5723,
+      "step": 5057
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.141797197032151e-05,
+      "loss": 3.8969,
+      "step": 5058
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.121187139323991e-05,
+      "loss": 4.2677,
+      "step": 5059
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.100577081615829e-05,
+      "loss": 4.1978,
+      "step": 5060
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.079967023907667e-05,
+      "loss": 4.1482,
+      "step": 5061
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.059356966199506e-05,
+      "loss": 4.096,
+      "step": 5062
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.038746908491344e-05,
+      "loss": 3.9407,
+      "step": 5063
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.0181368507831826e-05,
+      "loss": 3.9227,
+      "step": 5064
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.9975267930750205e-05,
+      "loss": 3.8739,
+      "step": 5065
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.976916735366859e-05,
+      "loss": 3.64,
+      "step": 5066
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.956306677658697e-05,
+      "loss": 3.4494,
+      "step": 5067
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.9356966199505365e-05,
+      "loss": 3.8111,
+      "step": 5068
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.9150865622423744e-05,
+      "loss": 4.1264,
+      "step": 5069
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.894476504534213e-05,
+      "loss": 4.3918,
+      "step": 5070
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.873866446826051e-05,
+      "loss": 4.2349,
+      "step": 5071
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.8532563891178896e-05,
+      "loss": 4.1164,
+      "step": 5072
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.8326463314097276e-05,
+      "loss": 4.1013,
+      "step": 5073
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.812036273701567e-05,
+      "loss": 4.354,
+      "step": 5074
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.791426215993405e-05,
+      "loss": 4.0763,
+      "step": 5075
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.7708161582852435e-05,
+      "loss": 4.1042,
+      "step": 5076
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.7502061005770815e-05,
+      "loss": 4.3442,
+      "step": 5077
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.72959604286892e-05,
+      "loss": 4.3586,
+      "step": 5078
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.708985985160758e-05,
+      "loss": 4.1992,
+      "step": 5079
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.6883759274525974e-05,
+      "loss": 4.2703,
+      "step": 5080
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.667765869744435e-05,
+      "loss": 4.1677,
+      "step": 5081
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.647155812036274e-05,
+      "loss": 3.7454,
+      "step": 5082
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.626545754328112e-05,
+      "loss": 4.2451,
+      "step": 5083
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.6059356966199506e-05,
+      "loss": 4.2227,
+      "step": 5084
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.585325638911789e-05,
+      "loss": 4.0489,
+      "step": 5085
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.564715581203628e-05,
+      "loss": 4.1475,
+      "step": 5086
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.544105523495466e-05,
+      "loss": 3.9289,
+      "step": 5087
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.5234954657873044e-05,
+      "loss": 4.256,
+      "step": 5088
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.5028854080791424e-05,
+      "loss": 3.8889,
+      "step": 5089
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.482275350370981e-05,
+      "loss": 4.1184,
+      "step": 5090
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.46166529266282e-05,
+      "loss": 3.9331,
+      "step": 5091
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.441055234954658e-05,
+      "loss": 4.2076,
+      "step": 5092
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.420445177246496e-05,
+      "loss": 4.0466,
+      "step": 5093
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.399835119538335e-05,
+      "loss": 4.0753,
+      "step": 5094
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.379225061830173e-05,
+      "loss": 3.8555,
+      "step": 5095
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.3586150041220115e-05,
+      "loss": 4.1324,
+      "step": 5096
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.33800494641385e-05,
+      "loss": 3.953,
+      "step": 5097
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.317394888705689e-05,
+      "loss": 4.03,
+      "step": 5098
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.296784830997527e-05,
+      "loss": 4.1562,
+      "step": 5099
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.2761747732893654e-05,
+      "loss": 3.9633,
+      "step": 5100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.255564715581203e-05,
+      "loss": 4.0636,
+      "step": 5101
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.234954657873042e-05,
+      "loss": 3.9551,
+      "step": 5102
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.2143446001648806e-05,
+      "loss": 3.9306,
+      "step": 5103
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.193734542456719e-05,
+      "loss": 3.8569,
+      "step": 5104
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.173124484748557e-05,
+      "loss": 3.9558,
+      "step": 5105
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.152514427040396e-05,
+      "loss": 3.9701,
+      "step": 5106
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.131904369332234e-05,
+      "loss": 3.8864,
+      "step": 5107
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.1112943116240724e-05,
+      "loss": 4.096,
+      "step": 5108
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.090684253915911e-05,
+      "loss": 3.9693,
+      "step": 5109
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.07007419620775e-05,
+      "loss": 3.8819,
+      "step": 5110
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.0494641384995876e-05,
+      "loss": 3.8667,
+      "step": 5111
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.028854080791426e-05,
+      "loss": 4.0236,
+      "step": 5112
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.008244023083264e-05,
+      "loss": 3.7302,
+      "step": 5113
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.9876339653751036e-05,
+      "loss": 3.9244,
+      "step": 5114
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.9670239076669415e-05,
+      "loss": 3.9832,
+      "step": 5115
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.94641384995878e-05,
+      "loss": 4.3451,
+      "step": 5116
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.925803792250618e-05,
+      "loss": 3.9485,
+      "step": 5117
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.905193734542457e-05,
+      "loss": 3.2527,
+      "step": 5118
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.884583676834295e-05,
+      "loss": 4.5447,
+      "step": 5119
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.863973619126134e-05,
+      "loss": 4.3387,
+      "step": 5120
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.843363561417972e-05,
+      "loss": 3.9065,
+      "step": 5121
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.8227535037098106e-05,
+      "loss": 4.2356,
+      "step": 5122
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.8021434460016486e-05,
+      "loss": 4.1724,
+      "step": 5123
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.781533388293487e-05,
+      "loss": 4.2051,
+      "step": 5124
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.760923330585325e-05,
+      "loss": 4.3791,
+      "step": 5125
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.7403132728771645e-05,
+      "loss": 4.3519,
+      "step": 5126
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.7197032151690024e-05,
+      "loss": 4.0761,
+      "step": 5127
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.699093157460841e-05,
+      "loss": 4.2142,
+      "step": 5128
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.678483099752679e-05,
+      "loss": 4.2152,
+      "step": 5129
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.657873042044518e-05,
+      "loss": 4.129,
+      "step": 5130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.6372629843363556e-05,
+      "loss": 4.3452,
+      "step": 5131
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.616652926628195e-05,
+      "loss": 3.9696,
+      "step": 5132
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.596042868920033e-05,
+      "loss": 4.1646,
+      "step": 5133
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5754328112118715e-05,
+      "loss": 3.9093,
+      "step": 5134
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5548227535037095e-05,
+      "loss": 4.2044,
+      "step": 5135
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.534212695795548e-05,
+      "loss": 4.0524,
+      "step": 5136
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.513602638087387e-05,
+      "loss": 4.1241,
+      "step": 5137
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.4929925803792254e-05,
+      "loss": 4.027,
+      "step": 5138
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.472382522671064e-05,
+      "loss": 3.9092,
+      "step": 5139
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.451772464962902e-05,
+      "loss": 3.9835,
+      "step": 5140
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.4311624072547406e-05,
+      "loss": 4.3999,
+      "step": 5141
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.4105523495465786e-05,
+      "loss": 4.4517,
+      "step": 5142
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.389942291838417e-05,
+      "loss": 4.4527,
+      "step": 5143
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.369332234130256e-05,
+      "loss": 4.0469,
+      "step": 5144
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.3487221764220945e-05,
+      "loss": 3.9777,
+      "step": 5145
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.3281121187139325e-05,
+      "loss": 3.9578,
+      "step": 5146
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.307502061005771e-05,
+      "loss": 4.2214,
+      "step": 5147
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.286892003297609e-05,
+      "loss": 3.9714,
+      "step": 5148
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.2662819455894484e-05,
+      "loss": 3.9987,
+      "step": 5149
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.245671887881286e-05,
+      "loss": 4.2112,
+      "step": 5150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.225061830173125e-05,
+      "loss": 3.9002,
+      "step": 5151
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.204451772464963e-05,
+      "loss": 4.0008,
+      "step": 5152
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.1838417147568016e-05,
+      "loss": 4.3579,
+      "step": 5153
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.1632316570486395e-05,
+      "loss": 4.3243,
+      "step": 5154
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.142621599340479e-05,
+      "loss": 4.0246,
+      "step": 5155
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.122011541632317e-05,
+      "loss": 3.8026,
+      "step": 5156
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.1014014839241554e-05,
+      "loss": 3.9824,
+      "step": 5157
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.0807914262159934e-05,
+      "loss": 4.6415,
+      "step": 5158
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.060181368507832e-05,
+      "loss": 4.1666,
+      "step": 5159
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.03957131079967e-05,
+      "loss": 4.2007,
+      "step": 5160
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.018961253091509e-05,
+      "loss": 4.2395,
+      "step": 5161
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.998351195383347e-05,
+      "loss": 4.1297,
+      "step": 5162
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.977741137675186e-05,
+      "loss": 3.7826,
+      "step": 5163
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.957131079967024e-05,
+      "loss": 3.9455,
+      "step": 5164
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.9365210222588625e-05,
+      "loss": 4.062,
+      "step": 5165
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.9159109645507004e-05,
+      "loss": 3.8099,
+      "step": 5166
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.89530090684254e-05,
+      "loss": 3.5225,
+      "step": 5167
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.874690849134378e-05,
+      "loss": 3.834,
+      "step": 5168
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.8540807914262164e-05,
+      "loss": 4.4107,
+      "step": 5169
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.833470733718054e-05,
+      "loss": 4.0043,
+      "step": 5170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.812860676009893e-05,
+      "loss": 4.0629,
+      "step": 5171
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.792250618301731e-05,
+      "loss": 4.2326,
+      "step": 5172
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.77164056059357e-05,
+      "loss": 3.9515,
+      "step": 5173
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.751030502885408e-05,
+      "loss": 4.2363,
+      "step": 5174
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.730420445177247e-05,
+      "loss": 3.9657,
+      "step": 5175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.709810387469085e-05,
+      "loss": 4.3529,
+      "step": 5176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.6892003297609234e-05,
+      "loss": 4.2016,
+      "step": 5177
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.6685902720527614e-05,
+      "loss": 4.359,
+      "step": 5178
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.647980214344601e-05,
+      "loss": 4.001,
+      "step": 5179
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.6273701566364386e-05,
+      "loss": 4.3112,
+      "step": 5180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.606760098928277e-05,
+      "loss": 4.1026,
+      "step": 5181
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.586150041220115e-05,
+      "loss": 4.3664,
+      "step": 5182
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.565539983511954e-05,
+      "loss": 3.9471,
+      "step": 5183
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5449299258037925e-05,
+      "loss": 3.8268,
+      "step": 5184
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.524319868095631e-05,
+      "loss": 4.0782,
+      "step": 5185
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.503709810387469e-05,
+      "loss": 4.1579,
+      "step": 5186
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.483099752679308e-05,
+      "loss": 4.283,
+      "step": 5187
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.462489694971146e-05,
+      "loss": 3.8805,
+      "step": 5188
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.441879637262984e-05,
+      "loss": 3.9799,
+      "step": 5189
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.421269579554823e-05,
+      "loss": 4.0569,
+      "step": 5190
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.4006595218466616e-05,
+      "loss": 4.0855,
+      "step": 5191
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3800494641384996e-05,
+      "loss": 4.0263,
+      "step": 5192
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.359439406430338e-05,
+      "loss": 3.9441,
+      "step": 5193
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.338829348722176e-05,
+      "loss": 4.2042,
+      "step": 5194
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.318219291014015e-05,
+      "loss": 4.2335,
+      "step": 5195
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2976092333058534e-05,
+      "loss": 3.9642,
+      "step": 5196
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.276999175597692e-05,
+      "loss": 4.1121,
+      "step": 5197
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.25638911788953e-05,
+      "loss": 3.9967,
+      "step": 5198
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.235779060181369e-05,
+      "loss": 4.2168,
+      "step": 5199
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2151690024732066e-05,
+      "loss": 4.2705,
+      "step": 5200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.194558944765045e-05,
+      "loss": 4.2795,
+      "step": 5201
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.173948887056884e-05,
+      "loss": 3.8897,
+      "step": 5202
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1533388293487225e-05,
+      "loss": 4.0218,
+      "step": 5203
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1327287716405605e-05,
+      "loss": 3.8457,
+      "step": 5204
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.112118713932399e-05,
+      "loss": 3.959,
+      "step": 5205
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.091508656224238e-05,
+      "loss": 3.8227,
+      "step": 5206
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.070898598516076e-05,
+      "loss": 3.8227,
+      "step": 5207
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.0502885408079144e-05,
+      "loss": 4.1956,
+      "step": 5208
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.029678483099753e-05,
+      "loss": 4.2759,
+      "step": 5209
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.0090684253915913e-05,
+      "loss": 3.9002,
+      "step": 5210
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9884583676834296e-05,
+      "loss": 3.6254,
+      "step": 5211
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9678483099752682e-05,
+      "loss": 4.0424,
+      "step": 5212
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9472382522671065e-05,
+      "loss": 3.4788,
+      "step": 5213
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9266281945589448e-05,
+      "loss": 3.8729,
+      "step": 5214
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9060181368507835e-05,
+      "loss": 4.2208,
+      "step": 5215
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.8854080791426218e-05,
+      "loss": 3.5773,
+      "step": 5216
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.86479802143446e-05,
+      "loss": 3.7965,
+      "step": 5217
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.8441879637262987e-05,
+      "loss": 3.1802,
+      "step": 5218
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.823577906018137e-05,
+      "loss": 4.32,
+      "step": 5219
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8029678483099753e-05,
+      "loss": 4.3693,
+      "step": 5220
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.782357790601814e-05,
+      "loss": 4.2193,
+      "step": 5221
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7617477328936522e-05,
+      "loss": 4.0533,
+      "step": 5222
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7411376751854905e-05,
+      "loss": 4.4829,
+      "step": 5223
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.720527617477329e-05,
+      "loss": 4.3755,
+      "step": 5224
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6999175597691674e-05,
+      "loss": 4.6229,
+      "step": 5225
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6793075020610057e-05,
+      "loss": 4.1862,
+      "step": 5226
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6586974443528444e-05,
+      "loss": 4.5177,
+      "step": 5227
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6380873866446827e-05,
+      "loss": 3.9135,
+      "step": 5228
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.617477328936521e-05,
+      "loss": 4.1186,
+      "step": 5229
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5968672712283596e-05,
+      "loss": 4.1583,
+      "step": 5230
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.576257213520198e-05,
+      "loss": 3.9471,
+      "step": 5231
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5556471558120362e-05,
+      "loss": 4.2787,
+      "step": 5232
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.535037098103875e-05,
+      "loss": 4.1871,
+      "step": 5233
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.514427040395713e-05,
+      "loss": 3.7932,
+      "step": 5234
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4938169826875518e-05,
+      "loss": 4.1597,
+      "step": 5235
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.47320692497939e-05,
+      "loss": 4.3627,
+      "step": 5236
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4525968672712284e-05,
+      "loss": 3.7963,
+      "step": 5237
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.431986809563067e-05,
+      "loss": 4.3078,
+      "step": 5238
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4113767518549053e-05,
+      "loss": 3.8685,
+      "step": 5239
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3907666941467436e-05,
+      "loss": 4.0759,
+      "step": 5240
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3701566364385822e-05,
+      "loss": 4.204,
+      "step": 5241
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3495465787304205e-05,
+      "loss": 3.8131,
+      "step": 5242
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.328936521022259e-05,
+      "loss": 4.0814,
+      "step": 5243
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3083264633140975e-05,
+      "loss": 4.1443,
+      "step": 5244
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2877164056059358e-05,
+      "loss": 3.8772,
+      "step": 5245
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.267106347897774e-05,
+      "loss": 3.8887,
+      "step": 5246
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2464962901896127e-05,
+      "loss": 4.2044,
+      "step": 5247
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.225886232481451e-05,
+      "loss": 4.0238,
+      "step": 5248
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2052761747732893e-05,
+      "loss": 4.0893,
+      "step": 5249
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.184666117065128e-05,
+      "loss": 4.0771,
+      "step": 5250
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1640560593569662e-05,
+      "loss": 3.8143,
+      "step": 5251
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1434460016488045e-05,
+      "loss": 3.9975,
+      "step": 5252
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.122835943940643e-05,
+      "loss": 4.0832,
+      "step": 5253
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1022258862324815e-05,
+      "loss": 3.9542,
+      "step": 5254
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0816158285243198e-05,
+      "loss": 3.8595,
+      "step": 5255
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0610057708161584e-05,
+      "loss": 4.1355,
+      "step": 5256
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0403957131079967e-05,
+      "loss": 4.0424,
+      "step": 5257
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.019785655399835e-05,
+      "loss": 3.8558,
+      "step": 5258
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9991755976916736e-05,
+      "loss": 3.9654,
+      "step": 5259
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.978565539983512e-05,
+      "loss": 4.126,
+      "step": 5260
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9579554822753502e-05,
+      "loss": 3.7017,
+      "step": 5261
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.937345424567189e-05,
+      "loss": 4.0207,
+      "step": 5262
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.916735366859027e-05,
+      "loss": 3.9261,
+      "step": 5263
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8961253091508655e-05,
+      "loss": 3.6222,
+      "step": 5264
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.875515251442704e-05,
+      "loss": 3.4982,
+      "step": 5265
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8549051937345424e-05,
+      "loss": 3.691,
+      "step": 5266
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8342951360263807e-05,
+      "loss": 3.6158,
+      "step": 5267
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8136850783182193e-05,
+      "loss": 3.542,
+      "step": 5268
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7930750206100576e-05,
+      "loss": 4.5392,
+      "step": 5269
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7724649629018963e-05,
+      "loss": 4.1642,
+      "step": 5270
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7518549051937346e-05,
+      "loss": 4.2862,
+      "step": 5271
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.731244847485573e-05,
+      "loss": 4.3042,
+      "step": 5272
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7106347897774115e-05,
+      "loss": 3.9891,
+      "step": 5273
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6900247320692498e-05,
+      "loss": 4.2369,
+      "step": 5274
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.669414674361088e-05,
+      "loss": 4.1768,
+      "step": 5275
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6488046166529267e-05,
+      "loss": 4.0495,
+      "step": 5276
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.628194558944765e-05,
+      "loss": 4.5904,
+      "step": 5277
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6075845012366033e-05,
+      "loss": 4.2252,
+      "step": 5278
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.586974443528442e-05,
+      "loss": 3.9242,
+      "step": 5279
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5663643858202802e-05,
+      "loss": 3.7498,
+      "step": 5280
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.545754328112119e-05,
+      "loss": 4.0022,
+      "step": 5281
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5251442704039572e-05,
+      "loss": 4.1743,
+      "step": 5282
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5045342126957956e-05,
+      "loss": 4.275,
+      "step": 5283
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4839241549876341e-05,
+      "loss": 4.3364,
+      "step": 5284
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4633140972794724e-05,
+      "loss": 4.4228,
+      "step": 5285
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4427040395713109e-05,
+      "loss": 4.2602,
+      "step": 5286
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4220939818631493e-05,
+      "loss": 4.1251,
+      "step": 5287
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4014839241549876e-05,
+      "loss": 3.9478,
+      "step": 5288
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3808738664468261e-05,
+      "loss": 4.029,
+      "step": 5289
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3602638087386646e-05,
+      "loss": 4.1337,
+      "step": 5290
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3396537510305029e-05,
+      "loss": 3.9954,
+      "step": 5291
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3190436933223413e-05,
+      "loss": 4.002,
+      "step": 5292
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2984336356141798e-05,
+      "loss": 4.1189,
+      "step": 5293
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2778235779060181e-05,
+      "loss": 4.1794,
+      "step": 5294
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2572135201978566e-05,
+      "loss": 4.3615,
+      "step": 5295
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.236603462489695e-05,
+      "loss": 4.0728,
+      "step": 5296
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2159934047815335e-05,
+      "loss": 3.9526,
+      "step": 5297
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1953833470733718e-05,
+      "loss": 4.3571,
+      "step": 5298
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1747732893652103e-05,
+      "loss": 4.0555,
+      "step": 5299
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1541632316570487e-05,
+      "loss": 4.1111,
+      "step": 5300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.133553173948887e-05,
+      "loss": 4.4715,
+      "step": 5301
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1129431162407255e-05,
+      "loss": 4.055,
+      "step": 5302
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.092333058532564e-05,
+      "loss": 4.1508,
+      "step": 5303
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0717230008244023e-05,
+      "loss": 4.1274,
+      "step": 5304
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0511129431162407e-05,
+      "loss": 3.7067,
+      "step": 5305
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0305028854080792e-05,
+      "loss": 3.772,
+      "step": 5306
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0098928276999175e-05,
+      "loss": 4.1527,
+      "step": 5307
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.89282769991756e-06,
+      "loss": 4.0173,
+      "step": 5308
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.686727122835944e-06,
+      "loss": 3.813,
+      "step": 5309
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.480626545754327e-06,
+      "loss": 4.0848,
+      "step": 5310
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.274525968672712e-06,
+      "loss": 3.9493,
+      "step": 5311
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.068425391591097e-06,
+      "loss": 3.984,
+      "step": 5312
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.862324814509481e-06,
+      "loss": 4.0913,
+      "step": 5313
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.656224237427864e-06,
+      "loss": 3.7076,
+      "step": 5314
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.450123660346249e-06,
+      "loss": 3.6559,
+      "step": 5315
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.244023083264634e-06,
+      "loss": 3.6492,
+      "step": 5316
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.037922506183017e-06,
+      "loss": 3.7781,
+      "step": 5317
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.831821929101401e-06,
+      "loss": 3.4142,
+      "step": 5318
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.625721352019786e-06,
+      "loss": 4.2849,
+      "step": 5319
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.4196207749381706e-06,
+      "loss": 4.1332,
+      "step": 5320
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.213520197856554e-06,
+      "loss": 3.9692,
+      "step": 5321
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.007419620774938e-06,
+      "loss": 4.0988,
+      "step": 5322
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.801319043693323e-06,
+      "loss": 4.1455,
+      "step": 5323
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.595218466611707e-06,
+      "loss": 4.1699,
+      "step": 5324
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.3891178895300905e-06,
+      "loss": 4.132,
+      "step": 5325
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.183017312448475e-06,
+      "loss": 4.0142,
+      "step": 5326
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.976916735366859e-06,
+      "loss": 4.1243,
+      "step": 5327
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.770816158285244e-06,
+      "loss": 4.1452,
+      "step": 5328
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.5647155812036275e-06,
+      "loss": 4.2982,
+      "step": 5329
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.358615004122011e-06,
+      "loss": 4.1401,
+      "step": 5330
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.152514427040396e-06,
+      "loss": 4.2774,
+      "step": 5331
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.94641384995878e-06,
+      "loss": 4.0217,
+      "step": 5332
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.740313272877164e-06,
+      "loss": 3.9812,
+      "step": 5333
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.534212695795548e-06,
+      "loss": 3.8157,
+      "step": 5334
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.328112118713932e-06,
+      "loss": 4.0374,
+      "step": 5335
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.122011541632317e-06,
+      "loss": 3.9363,
+      "step": 5336
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.915910964550701e-06,
+      "loss": 4.2199,
+      "step": 5337
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7098103874690853e-06,
+      "loss": 4.1038,
+      "step": 5338
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.503709810387469e-06,
+      "loss": 4.1038,
+      "step": 5339
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2976092333058533e-06,
+      "loss": 3.9915,
+      "step": 5340
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.0915086562242376e-06,
+      "loss": 4.0858,
+      "step": 5341
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.885408079142622e-06,
+      "loss": 3.8158,
+      "step": 5342
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6793075020610057e-06,
+      "loss": 3.6785,
+      "step": 5343
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.47320692497939e-06,
+      "loss": 4.128,
+      "step": 5344
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.267106347897774e-06,
+      "loss": 4.0782,
+      "step": 5345
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0610057708161584e-06,
+      "loss": 3.9853,
+      "step": 5346
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8549051937345426e-06,
+      "loss": 4.0732,
+      "step": 5347
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.6488046166529267e-06,
+      "loss": 4.041,
+      "step": 5348
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.442704039571311e-06,
+      "loss": 3.915,
+      "step": 5349
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.236603462489695e-06,
+      "loss": 3.8774,
+      "step": 5350
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0305028854080792e-06,
+      "loss": 3.4855,
+      "step": 5351
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.244023083264633e-07,
+      "loss": 3.2947,
+      "step": 5352
+    },
+    {
+      "epoch": 3.0,
+      "step": 5352,
+      "total_flos": 0.0,
+      "train_loss": 5.036994248450426,
+      "train_runtime": 29700.184,
+      "train_samples_per_second": 2.883,
+      "train_steps_per_second": 0.18
+    }
+  ],
+  "max_steps": 5352,
+  "num_train_epochs": 3,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}