diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,47940 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 7985,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333334e-08,
+      "loss": 4.2336,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666668e-07,
+      "loss": 4.3981,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 4.2501,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3333333333333335e-07,
+      "loss": 4.3387,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1666666666666667e-07,
+      "loss": 4.3037,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 4.3054,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.833333333333334e-07,
+      "loss": 4.3313,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 4.4159,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.5e-07,
+      "loss": 4.4125,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 4.2751,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.166666666666666e-07,
+      "loss": 4.3363,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 4.134,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0833333333333335e-06,
+      "loss": 4.0503,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1666666666666668e-06,
+      "loss": 4.1246,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-06,
+      "loss": 4.0457,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 3.9535,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4166666666666667e-06,
+      "loss": 3.9248,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 3.7212,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5833333333333333e-06,
+      "loss": 3.739,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 3.8158,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.75e-06,
+      "loss": 3.6857,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8333333333333333e-06,
+      "loss": 3.3812,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.916666666666667e-06,
+      "loss": 3.2823,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 3.3037,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 3.1563,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.166666666666667e-06,
+      "loss": 3.2088,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.25e-06,
+      "loss": 3.2445,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3333333333333336e-06,
+      "loss": 3.0857,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4166666666666667e-06,
+      "loss": 3.1885,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 3.0821,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5833333333333337e-06,
+      "loss": 3.041,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 3.0185,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 2.9155,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8333333333333335e-06,
+      "loss": 2.9572,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.916666666666667e-06,
+      "loss": 2.9639,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 2.903,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0833333333333336e-06,
+      "loss": 2.8959,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1666666666666667e-06,
+      "loss": 2.8931,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 2.8943,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 2.8628,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.416666666666667e-06,
+      "loss": 2.8585,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-06,
+      "loss": 2.9245,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5833333333333335e-06,
+      "loss": 2.8505,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6666666666666666e-06,
+      "loss": 2.7743,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 2.7836,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.833333333333334e-06,
+      "loss": 2.8941,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.916666666666667e-06,
+      "loss": 2.8115,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.8246,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.083333333333334e-06,
+      "loss": 2.7951,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 2.8162,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.25e-06,
+      "loss": 2.846,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.333333333333334e-06,
+      "loss": 2.744,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.416666666666667e-06,
+      "loss": 2.8203,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-06,
+      "loss": 2.7928,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.583333333333333e-06,
+      "loss": 2.829,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 2.776,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.75e-06,
+      "loss": 2.7287,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.833333333333333e-06,
+      "loss": 2.7728,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9166666666666665e-06,
+      "loss": 2.74,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 2.7734,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.0833333333333335e-06,
+      "loss": 2.6793,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1666666666666675e-06,
+      "loss": 2.6302,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2500000000000006e-06,
+      "loss": 2.8271,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 2.6797,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.416666666666667e-06,
+      "loss": 2.7131,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 2.732,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.583333333333334e-06,
+      "loss": 2.8437,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.666666666666667e-06,
+      "loss": 2.765,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.75e-06,
+      "loss": 2.6763,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 2.7606,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.916666666666667e-06,
+      "loss": 2.7203,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 2.7052,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.083333333333333e-06,
+      "loss": 2.7016,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.166666666666667e-06,
+      "loss": 2.7441,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.25e-06,
+      "loss": 2.6667,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.333333333333333e-06,
+      "loss": 2.678,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.416666666666667e-06,
+      "loss": 2.6615,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 2.829,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5833333333333335e-06,
+      "loss": 2.7254,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 2.7096,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 2.7907,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.833333333333334e-06,
+      "loss": 2.6845,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.916666666666667e-06,
+      "loss": 2.7505,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7e-06,
+      "loss": 2.7554,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.083333333333335e-06,
+      "loss": 2.7679,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.166666666666667e-06,
+      "loss": 2.6603,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.25e-06,
+      "loss": 2.8005,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 2.704,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.416666666666668e-06,
+      "loss": 2.7141,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.7597,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.583333333333333e-06,
+      "loss": 2.7011,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.666666666666667e-06,
+      "loss": 2.6519,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.75e-06,
+      "loss": 2.6721,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.833333333333333e-06,
+      "loss": 2.7677,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.916666666666667e-06,
+      "loss": 2.6248,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.6609,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.083333333333334e-06,
+      "loss": 2.7138,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.166666666666668e-06,
+      "loss": 2.7016,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.25e-06,
+      "loss": 2.675,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 2.8009,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.416666666666667e-06,
+      "loss": 2.7373,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.5e-06,
+      "loss": 2.65,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.583333333333333e-06,
+      "loss": 2.6697,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 2.6336,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 2.5617,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.833333333333334e-06,
+      "loss": 2.7599,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.916666666666667e-06,
+      "loss": 2.5874,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-06,
+      "loss": 2.8053,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.083333333333333e-06,
+      "loss": 2.6765,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 2.7416,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.250000000000001e-06,
+      "loss": 2.7031,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 2.6636,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.416666666666667e-06,
+      "loss": 2.6386,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.5e-06,
+      "loss": 2.6176,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.583333333333335e-06,
+      "loss": 2.6059,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.666666666666667e-06,
+      "loss": 2.6198,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.75e-06,
+      "loss": 2.6881,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.833333333333333e-06,
+      "loss": 2.7184,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.916666666666668e-06,
+      "loss": 2.6249,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 2.711,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0083333333333334e-05,
+      "loss": 2.6357,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0166666666666667e-05,
+      "loss": 2.7275,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.025e-05,
+      "loss": 2.6683,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0333333333333335e-05,
+      "loss": 2.6425,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0416666666666668e-05,
+      "loss": 2.6689,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0500000000000001e-05,
+      "loss": 2.6011,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0583333333333334e-05,
+      "loss": 2.7151,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 2.5622,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.075e-05,
+      "loss": 2.5775,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 2.5882,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0916666666666667e-05,
+      "loss": 2.7168,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 2.6118,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1083333333333335e-05,
+      "loss": 2.679,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1166666666666668e-05,
+      "loss": 2.6089,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.125e-05,
+      "loss": 2.5694,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1333333333333334e-05,
+      "loss": 2.7026,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1416666666666667e-05,
+      "loss": 2.5866,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.15e-05,
+      "loss": 2.6475,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1583333333333335e-05,
+      "loss": 2.7063,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 2.6096,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1750000000000001e-05,
+      "loss": 2.7588,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1833333333333334e-05,
+      "loss": 2.597,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1916666666666667e-05,
+      "loss": 2.673,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 2.6224,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2083333333333333e-05,
+      "loss": 2.6218,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2166666666666667e-05,
+      "loss": 2.6406,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2250000000000001e-05,
+      "loss": 2.6584,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2333333333333334e-05,
+      "loss": 2.6789,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2416666666666667e-05,
+      "loss": 2.6636,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 2.5331,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2583333333333334e-05,
+      "loss": 2.65,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2666666666666667e-05,
+      "loss": 2.6706,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.275e-05,
+      "loss": 2.552,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2833333333333335e-05,
+      "loss": 2.552,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2916666666666668e-05,
+      "loss": 2.7169,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 2.6293,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3083333333333334e-05,
+      "loss": 2.5663,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3166666666666667e-05,
+      "loss": 2.6813,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.325e-05,
+      "loss": 2.6824,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 2.7004,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3416666666666666e-05,
+      "loss": 2.6353,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 2.6873,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3583333333333334e-05,
+      "loss": 2.6126,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3666666666666667e-05,
+      "loss": 2.6399,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.375e-05,
+      "loss": 2.6535,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3833333333333334e-05,
+      "loss": 2.5966,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3916666666666667e-05,
+      "loss": 2.6718,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4e-05,
+      "loss": 2.6614,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4083333333333336e-05,
+      "loss": 2.7528,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 2.6539,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.425e-05,
+      "loss": 2.6291,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4333333333333334e-05,
+      "loss": 2.6638,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4416666666666667e-05,
+      "loss": 2.5688,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.45e-05,
+      "loss": 2.6528,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4583333333333333e-05,
+      "loss": 2.637,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 2.6695,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4750000000000003e-05,
+      "loss": 2.6816,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4833333333333336e-05,
+      "loss": 2.6676,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4916666666666669e-05,
+      "loss": 2.6553,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 2.6288,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5083333333333333e-05,
+      "loss": 2.6088,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5166666666666667e-05,
+      "loss": 2.5899,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.525e-05,
+      "loss": 2.5588,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 2.7989,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5416666666666668e-05,
+      "loss": 2.7723,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.55e-05,
+      "loss": 2.6612,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5583333333333334e-05,
+      "loss": 2.6192,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5666666666666667e-05,
+      "loss": 2.6141,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.575e-05,
+      "loss": 2.6528,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 2.6614,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5916666666666666e-05,
+      "loss": 2.6158,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 2.7284,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6083333333333336e-05,
+      "loss": 2.6423,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.616666666666667e-05,
+      "loss": 2.5977,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 2.6677,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6333333333333335e-05,
+      "loss": 2.6983,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6416666666666668e-05,
+      "loss": 2.625,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.65e-05,
+      "loss": 2.7025,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6583333333333334e-05,
+      "loss": 2.5715,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 2.6548,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.675e-05,
+      "loss": 2.6732,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6833333333333334e-05,
+      "loss": 2.6652,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6916666666666667e-05,
+      "loss": 2.7047,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7e-05,
+      "loss": 2.5851,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7083333333333333e-05,
+      "loss": 2.6661,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7166666666666666e-05,
+      "loss": 2.6062,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7250000000000003e-05,
+      "loss": 2.6519,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 2.6822,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.741666666666667e-05,
+      "loss": 2.6287,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 2.599,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7583333333333335e-05,
+      "loss": 2.6425,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7666666666666668e-05,
+      "loss": 2.6907,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.775e-05,
+      "loss": 2.712,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7833333333333334e-05,
+      "loss": 2.6715,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7916666666666667e-05,
+      "loss": 2.6411,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8e-05,
+      "loss": 2.6661,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8083333333333334e-05,
+      "loss": 2.6475,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8166666666666667e-05,
+      "loss": 2.7041,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.825e-05,
+      "loss": 2.573,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 2.6374,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8416666666666666e-05,
+      "loss": 2.6451,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8500000000000002e-05,
+      "loss": 2.598,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8583333333333336e-05,
+      "loss": 2.5762,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 2.6392,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 2.5707,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8833333333333335e-05,
+      "loss": 2.6303,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8916666666666668e-05,
+      "loss": 2.6154,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9e-05,
+      "loss": 2.6441,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9083333333333338e-05,
+      "loss": 2.7096,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 2.5967,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.925e-05,
+      "loss": 2.6077,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9333333333333333e-05,
+      "loss": 2.6727,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9416666666666667e-05,
+      "loss": 2.6149,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.95e-05,
+      "loss": 2.7113,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9583333333333333e-05,
+      "loss": 2.7259,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9666666666666666e-05,
+      "loss": 2.5027,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9750000000000002e-05,
+      "loss": 2.5828,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9833333333333335e-05,
+      "loss": 2.5211,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.991666666666667e-05,
+      "loss": 2.583,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 2.7088,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999177327654e-05,
+      "loss": 2.6475,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999996709310747e-05,
+      "loss": 2.6637,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999992595949688e-05,
+      "loss": 2.6556,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998683724515e-05,
+      "loss": 2.7254,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997943319808e-05,
+      "loss": 2.5848,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999970383809707e-05,
+      "loss": 2.6536,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999959689081505e-05,
+      "loss": 2.6,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999947349015243e-05,
+      "loss": 2.726,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993336361295e-05,
+      "loss": 2.612,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999991773287693e-05,
+      "loss": 2.706,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999900456809742e-05,
+      "loss": 2.6978,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999988153541424e-05,
+      "loss": 2.6943,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999986096869354e-05,
+      "loss": 2.6946,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999838756651017e-05,
+      "loss": 2.61,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999814899290326e-05,
+      "loss": 2.5612,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99997893966154e-05,
+      "loss": 2.6238,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999976224863043e-05,
+      "loss": 2.652,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999973345533988e-05,
+      "loss": 2.7256,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999703016748497e-05,
+      "loss": 2.6758,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999670932861277e-05,
+      "loss": 2.6443,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999637203683508e-05,
+      "loss": 2.5538,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999601829220732e-05,
+      "loss": 2.6315,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999956480947878e-05,
+      "loss": 2.6319,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999526144463734e-05,
+      "loss": 2.6485,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999485834181963e-05,
+      "loss": 2.662,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999944387864009e-05,
+      "loss": 2.6503,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999400277845025e-05,
+      "loss": 2.6607,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999935503180394e-05,
+      "loss": 2.6542,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999930814052428e-05,
+      "loss": 2.6189,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999259604013758e-05,
+      "loss": 2.6097,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999209422280362e-05,
+      "loss": 2.6257,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999157595332352e-05,
+      "loss": 2.6093,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999910412317825e-05,
+      "loss": 2.693,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999049005826855e-05,
+      "loss": 2.6123,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998992243287233e-05,
+      "loss": 2.7042,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998933835568732e-05,
+      "loss": 2.6559,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998873782680953e-05,
+      "loss": 2.6979,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998812084633786e-05,
+      "loss": 2.6517,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999874874143737e-05,
+      "loss": 2.6751,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999868375310214e-05,
+      "loss": 2.5936,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998617119638777e-05,
+      "loss": 2.5678,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998548841058258e-05,
+      "loss": 2.6304,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998478917371805e-05,
+      "loss": 2.6805,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999840734859093e-05,
+      "loss": 2.5785,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998334134727405e-05,
+      "loss": 2.602,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998259275793276e-05,
+      "loss": 2.5824,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998182771800865e-05,
+      "loss": 2.6257,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999810462276275e-05,
+      "loss": 2.6301,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998024828691804e-05,
+      "loss": 2.6651,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999794338960114e-05,
+      "loss": 2.6086,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997860305504165e-05,
+      "loss": 2.6723,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999777557641455e-05,
+      "loss": 2.5765,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997689202346234e-05,
+      "loss": 2.6225,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999760118331343e-05,
+      "loss": 2.6989,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997511519330618e-05,
+      "loss": 2.6739,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999742021041255e-05,
+      "loss": 2.6199,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999732725657425e-05,
+      "loss": 2.6152,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997232657831018e-05,
+      "loss": 2.6659,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997136414198414e-05,
+      "loss": 2.6152,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999703852569227e-05,
+      "loss": 2.5664,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996938992328702e-05,
+      "loss": 2.7887,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996837814124073e-05,
+      "loss": 2.5899,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996734991095045e-05,
+      "loss": 2.6093,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996630523258523e-05,
+      "loss": 2.6652,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996524410631703e-05,
+      "loss": 2.7258,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996416653232044e-05,
+      "loss": 2.6488,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996307251077272e-05,
+      "loss": 2.6967,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999619620418539e-05,
+      "loss": 2.6332,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999608351257467e-05,
+      "loss": 2.6484,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999596917626365e-05,
+      "loss": 2.5999,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995853195271148e-05,
+      "loss": 2.5495,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995735569616238e-05,
+      "loss": 2.5845,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995616299318285e-05,
+      "loss": 2.6423,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995495384396902e-05,
+      "loss": 2.5772,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995372824871992e-05,
+      "loss": 2.5689,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995248620763714e-05,
+      "loss": 2.5596,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999512277209251e-05,
+      "loss": 2.7031,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999499527887908e-05,
+      "loss": 2.6386,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994866141144404e-05,
+      "loss": 2.5685,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994735358909733e-05,
+      "loss": 2.5881,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999460293219658e-05,
+      "loss": 2.5738,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994468861026737e-05,
+      "loss": 2.6579,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999433314542226e-05,
+      "loss": 2.6457,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994195785405485e-05,
+      "loss": 2.6021,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994056780999003e-05,
+      "loss": 2.5963,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993916132225694e-05,
+      "loss": 2.6122,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993773839108694e-05,
+      "loss": 2.6923,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993629901671418e-05,
+      "loss": 2.6092,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999348431993755e-05,
+      "loss": 2.5937,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999333709393104e-05,
+      "loss": 2.5972,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993188223676113e-05,
+      "loss": 2.6518,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993037709197262e-05,
+      "loss": 2.553,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999288555051925e-05,
+      "loss": 2.7012,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999273174766712e-05,
+      "loss": 2.6158,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992576300666173e-05,
+      "loss": 2.6712,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992419209541986e-05,
+      "loss": 2.6334,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99922604743204e-05,
+      "loss": 2.575,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992100095027543e-05,
+      "loss": 2.6051,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991938071689797e-05,
+      "loss": 2.6002,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999177440433382e-05,
+      "loss": 2.548,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999160909298654e-05,
+      "loss": 2.7554,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991442137675164e-05,
+      "loss": 2.6635,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991273538427154e-05,
+      "loss": 2.6372,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991103295270248e-05,
+      "loss": 2.5972,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999093140823247e-05,
+      "loss": 2.6956,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990757877342087e-05,
+      "loss": 2.5829,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999058270262766e-05,
+      "loss": 2.6409,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990405884118006e-05,
+      "loss": 2.6267,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990227421842223e-05,
+      "loss": 2.6865,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990047315829667e-05,
+      "loss": 2.651,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998986556610998e-05,
+      "loss": 2.592,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989682172713057e-05,
+      "loss": 2.5833,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998949713566908e-05,
+      "loss": 2.5785,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998931045500849e-05,
+      "loss": 2.5294,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989122130762007e-05,
+      "loss": 2.6863,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998893216296061e-05,
+      "loss": 2.7056,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998874055163556e-05,
+      "loss": 2.6358,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988547296818382e-05,
+      "loss": 2.7259,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988352398540872e-05,
+      "loss": 2.5349,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988155856835104e-05,
+      "loss": 2.6124,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987957671733405e-05,
+      "loss": 2.5836,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987757843268395e-05,
+      "loss": 2.668,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998755637147294e-05,
+      "loss": 2.6534,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987353256380203e-05,
+      "loss": 2.6116,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987148498023594e-05,
+      "loss": 2.6254,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986942096436803e-05,
+      "loss": 2.6446,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986734051653793e-05,
+      "loss": 2.675,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986524363708793e-05,
+      "loss": 2.6501,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998631303263631e-05,
+      "loss": 2.5856,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986100058471106e-05,
+      "loss": 2.5772,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985885441248224e-05,
+      "loss": 2.5674,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985669181002985e-05,
+      "loss": 2.68,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998545127777096e-05,
+      "loss": 2.5854,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998523173158801e-05,
+      "loss": 2.5075,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998501054249025e-05,
+      "loss": 2.6517,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998478771051408e-05,
+      "loss": 2.6698,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998456323569616e-05,
+      "loss": 2.5885,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984337118073424e-05,
+      "loss": 2.6414,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998410935768308e-05,
+      "loss": 2.5813,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983879954562597e-05,
+      "loss": 2.66,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983648908749724e-05,
+      "loss": 2.6351,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998341622028247e-05,
+      "loss": 2.6007,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983181889199132e-05,
+      "loss": 2.5747,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982945915538253e-05,
+      "loss": 2.6606,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982708299338665e-05,
+      "loss": 2.5565,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982469040639464e-05,
+      "loss": 2.6114,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982228139480014e-05,
+      "loss": 2.5617,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981985595899952e-05,
+      "loss": 2.6256,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981741409939187e-05,
+      "loss": 2.6255,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981495581637892e-05,
+      "loss": 2.6271,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981248111036518e-05,
+      "loss": 2.6216,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980998998175784e-05,
+      "loss": 2.5209,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998074824309667e-05,
+      "loss": 2.5854,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998049584584044e-05,
+      "loss": 2.6692,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980241806448625e-05,
+      "loss": 2.6134,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979986124963015e-05,
+      "loss": 2.6498,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979728801425682e-05,
+      "loss": 2.6719,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979469835878966e-05,
+      "loss": 2.6103,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979209228365474e-05,
+      "loss": 2.6735,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978946978928088e-05,
+      "loss": 2.6478,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978683087609955e-05,
+      "loss": 2.675,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997841755445449e-05,
+      "loss": 2.572,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997815037950539e-05,
+      "loss": 2.6396,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997788156280661e-05,
+      "loss": 2.6916,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977611104402377e-05,
+      "loss": 2.6106,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99773390043372e-05,
+      "loss": 2.639,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997706526265584e-05,
+      "loss": 2.6049,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997678987940334e-05,
+      "loss": 2.6178,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997651285462501e-05,
+      "loss": 2.6362,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997623418836643e-05,
+      "loss": 2.6161,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997595388067345e-05,
+      "loss": 2.7403,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975671931592192e-05,
+      "loss": 2.5724,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997538834116904e-05,
+      "loss": 2.638,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975103109450666e-05,
+      "loss": 2.6132,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997481623648399e-05,
+      "loss": 2.6978,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997452772231622e-05,
+      "loss": 2.6985,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997423756699482e-05,
+      "loss": 2.6719,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997394577056753e-05,
+      "loss": 2.6547,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997365233308237e-05,
+      "loss": 2.6419,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997335725458761e-05,
+      "loss": 2.6037,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997306053513181e-05,
+      "loss": 2.637,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997276217476378e-05,
+      "loss": 2.5824,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972462173532617e-05,
+      "loss": 2.539,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972160531487684e-05,
+      "loss": 2.5714,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971857248678607e-05,
+      "loss": 2.6026,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971552325155287e-05,
+      "loss": 2.6302,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971245760967897e-05,
+      "loss": 2.6532,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970937556166875e-05,
+      "loss": 2.6813,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997062771080293e-05,
+      "loss": 2.5767,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970316224927046e-05,
+      "loss": 2.568,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997000309859047e-05,
+      "loss": 2.604,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969688331844727e-05,
+      "loss": 2.6137,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99693719247416e-05,
+      "loss": 2.6715,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996905387733315e-05,
+      "loss": 2.5996,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996873418967171e-05,
+      "loss": 2.7046,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996841286180988e-05,
+      "loss": 2.5538,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996808989380053e-05,
+      "loss": 2.6429,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967765285696793e-05,
+      "loss": 2.5366,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967439037552086e-05,
+      "loss": 2.5471,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967111149420085e-05,
+      "loss": 2.5267,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996678162135474e-05,
+      "loss": 2.6466,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966450453410265e-05,
+      "loss": 2.5905,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966117645641152e-05,
+      "loss": 2.5116,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965783198102163e-05,
+      "loss": 2.5892,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965447110848317e-05,
+      "loss": 2.6033,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965109383934923e-05,
+      "loss": 2.6817,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964770017417535e-05,
+      "loss": 2.5417,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964429011352008e-05,
+      "loss": 2.6017,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964086365794433e-05,
+      "loss": 2.5886,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99637420808012e-05,
+      "loss": 2.6539,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963396156428945e-05,
+      "loss": 2.6638,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996304859273459e-05,
+      "loss": 2.5884,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996269938977532e-05,
+      "loss": 2.6443,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996234854760859e-05,
+      "loss": 2.5909,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996199606629213e-05,
+      "loss": 2.6505,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996164194588393e-05,
+      "loss": 2.5917,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961286186442258e-05,
+      "loss": 2.5543,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960928788025648e-05,
+      "loss": 2.5961,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960569750692906e-05,
+      "loss": 2.5508,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99602090745031e-05,
+      "loss": 2.5614,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995984675951558e-05,
+      "loss": 2.6346,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995948280578996e-05,
+      "loss": 2.5764,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959117213386115e-05,
+      "loss": 2.5601,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958749982364204e-05,
+      "loss": 2.6063,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958381112784653e-05,
+      "loss": 2.5831,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958010604708146e-05,
+      "loss": 2.6907,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957638458195643e-05,
+      "loss": 2.7066,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995726467330838e-05,
+      "loss": 2.4971,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995688925010786e-05,
+      "loss": 2.6276,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956512188655844e-05,
+      "loss": 2.6133,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995613348901438e-05,
+      "loss": 2.6331,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955753151245777e-05,
+      "loss": 2.628,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995537117541261e-05,
+      "loss": 2.7123,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954987561577727e-05,
+      "loss": 2.6864,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954602309804244e-05,
+      "loss": 2.5875,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954215420155556e-05,
+      "loss": 2.6417,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995382689269531e-05,
+      "loss": 2.6785,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995343672748744e-05,
+      "loss": 2.6286,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953044924596138e-05,
+      "loss": 2.654,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952651484085867e-05,
+      "loss": 2.68,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952256406021366e-05,
+      "loss": 2.5393,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951859690467637e-05,
+      "loss": 2.5672,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995146133748995e-05,
+      "loss": 2.6026,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995106134715386e-05,
+      "loss": 2.6212,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950659719525164e-05,
+      "loss": 2.5795,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995025645466995e-05,
+      "loss": 2.4936,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949851552654567e-05,
+      "loss": 2.5068,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994944501354564e-05,
+      "loss": 2.6199,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994903683741006e-05,
+      "loss": 2.6103,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948627024314976e-05,
+      "loss": 2.6668,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994821557432782e-05,
+      "loss": 2.5916,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947802487516302e-05,
+      "loss": 2.6213,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947387763948372e-05,
+      "loss": 2.6161,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946971403692275e-05,
+      "loss": 2.6469,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946553406816516e-05,
+      "loss": 2.568,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946133773389865e-05,
+      "loss": 2.592,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945712503481375e-05,
+      "loss": 2.6116,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945289597160352e-05,
+      "loss": 2.6831,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944865054496382e-05,
+      "loss": 2.6109,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944438875559315e-05,
+      "loss": 2.448,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944011060419278e-05,
+      "loss": 2.511,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994358160914665e-05,
+      "loss": 2.5483,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99431505218121e-05,
+      "loss": 2.5766,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942717798486556e-05,
+      "loss": 2.6464,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994228343924121e-05,
+      "loss": 2.6314,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941847444147534e-05,
+      "loss": 2.6027,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994140981327726e-05,
+      "loss": 2.5842,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940970546702402e-05,
+      "loss": 2.6051,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940529644495226e-05,
+      "loss": 2.5793,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994008710672828e-05,
+      "loss": 2.5876,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939642933474373e-05,
+      "loss": 2.6247,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993919712480659e-05,
+      "loss": 2.6165,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938749680798283e-05,
+      "loss": 2.5501,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938300601523065e-05,
+      "loss": 2.5619,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937849887054835e-05,
+      "loss": 2.6414,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937397537467743e-05,
+      "loss": 2.6265,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993694355283622e-05,
+      "loss": 2.6945,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993648793323496e-05,
+      "loss": 2.6384,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993603067873893e-05,
+      "loss": 2.6269,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935571789423363e-05,
+      "loss": 2.6714,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935111265363766e-05,
+      "loss": 2.6228,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934649106635903e-05,
+      "loss": 2.5338,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934185313315822e-05,
+      "loss": 2.6335,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993371988547983e-05,
+      "loss": 2.6301,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933252823204508e-05,
+      "loss": 2.4941,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932784126566698e-05,
+      "loss": 2.6192,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932313795643527e-05,
+      "loss": 2.6421,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931841830512373e-05,
+      "loss": 2.6386,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993136823125089e-05,
+      "loss": 2.5943,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930892997937007e-05,
+      "loss": 2.5378,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930416130648908e-05,
+      "loss": 2.5439,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929937629465062e-05,
+      "loss": 2.581,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99294574944642e-05,
+      "loss": 2.6202,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928975725725314e-05,
+      "loss": 2.6583,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992849232332767e-05,
+      "loss": 2.6032,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928007287350813e-05,
+      "loss": 2.5651,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927520617874546e-05,
+      "loss": 2.6047,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927032314978937e-05,
+      "loss": 2.5948,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926542378744333e-05,
+      "loss": 2.6075,
+      "step": 539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926050809251346e-05,
+      "loss": 2.6104,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925557606580853e-05,
+      "loss": 2.5165,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925062770814008e-05,
+      "loss": 2.5652,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924566302032223e-05,
+      "loss": 2.4964,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924068200317187e-05,
+      "loss": 2.6349,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923568465750854e-05,
+      "loss": 2.646,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992306709841545e-05,
+      "loss": 2.6406,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992256409839346e-05,
+      "loss": 2.618,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922059465767654e-05,
+      "loss": 2.5905,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992155320062106e-05,
+      "loss": 2.6743,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992104530303697e-05,
+      "loss": 2.5512,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920535773098954e-05,
+      "loss": 2.6035,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920024610890852e-05,
+      "loss": 2.6388,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919511816496757e-05,
+      "loss": 2.538,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918997390001052e-05,
+      "loss": 2.5902,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991848133148837e-05,
+      "loss": 2.6051,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991796364104363e-05,
+      "loss": 2.5522,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917444318752e-05,
+      "loss": 2.5221,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991692336469893e-05,
+      "loss": 2.6138,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916400778970134e-05,
+      "loss": 2.5753,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99158765616516e-05,
+      "loss": 2.5208,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915350712829575e-05,
+      "loss": 2.5198,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991482323259058e-05,
+      "loss": 2.6891,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99142941210214e-05,
+      "loss": 2.5855,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913763378209103e-05,
+      "loss": 2.6502,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991323100424101e-05,
+      "loss": 2.617,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912696999204706e-05,
+      "loss": 2.6272,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912161363188062e-05,
+      "loss": 2.5578,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911624096279206e-05,
+      "loss": 2.6116,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991108519856654e-05,
+      "loss": 2.6416,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910544670138728e-05,
+      "loss": 2.5591,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910002511084708e-05,
+      "loss": 2.6433,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990945872149368e-05,
+      "loss": 2.6009,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908913301455117e-05,
+      "loss": 2.5646,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908366251058764e-05,
+      "loss": 2.6412,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907817570394622e-05,
+      "loss": 2.6246,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907267259552977e-05,
+      "loss": 2.5621,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906715318624367e-05,
+      "loss": 2.6636,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906161747699607e-05,
+      "loss": 2.6397,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990560654686978e-05,
+      "loss": 2.6008,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905049716226236e-05,
+      "loss": 2.6805,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990449125586059e-05,
+      "loss": 2.6406,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990393116586473e-05,
+      "loss": 2.588,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990336944633081e-05,
+      "loss": 2.5496,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990280609735125e-05,
+      "loss": 2.5073,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902241119018747e-05,
+      "loss": 2.5559,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990167451142625e-05,
+      "loss": 2.5259,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901106274666992e-05,
+      "loss": 2.5887,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900536408834464e-05,
+      "loss": 2.5843,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899964914022432e-05,
+      "loss": 2.6002,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899391790324928e-05,
+      "loss": 2.7115,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898817037836243e-05,
+      "loss": 2.5289,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989824065665095e-05,
+      "loss": 2.5252,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989766264686388e-05,
+      "loss": 2.5621,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989708300857014e-05,
+      "loss": 2.5329,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896501741865094e-05,
+      "loss": 2.6227,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895918846844387e-05,
+      "loss": 2.5721,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895334323603922e-05,
+      "loss": 2.5717,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894748172239873e-05,
+      "loss": 2.5618,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894160392848684e-05,
+      "loss": 2.675,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893570985527063e-05,
+      "loss": 2.5626,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892979950371987e-05,
+      "loss": 2.4621,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892387287480703e-05,
+      "loss": 2.5161,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891792996950728e-05,
+      "loss": 2.6794,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891197078879837e-05,
+      "loss": 2.5395,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890599533366082e-05,
+      "loss": 2.661,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989000036050778e-05,
+      "loss": 2.7008,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889399560403517e-05,
+      "loss": 2.5962,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888797133152143e-05,
+      "loss": 2.5446,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988819307885278e-05,
+      "loss": 2.537,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887587397604814e-05,
+      "loss": 2.6199,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98869800895079e-05,
+      "loss": 2.5285,
+      "step": 611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886371154661962e-05,
+      "loss": 2.5873,
+      "step": 612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988576059316719e-05,
+      "loss": 2.6856,
+      "step": 613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885148405124046e-05,
+      "loss": 2.6341,
+      "step": 614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988453459063325e-05,
+      "loss": 2.6026,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98839191497958e-05,
+      "loss": 2.5801,
+      "step": 616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988330208271296e-05,
+      "loss": 2.5729,
+      "step": 617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988268338948625e-05,
+      "loss": 2.5951,
+      "step": 618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882063070217476e-05,
+      "loss": 2.5322,
+      "step": 619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881441125008695e-05,
+      "loss": 2.5562,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988081755396224e-05,
+      "loss": 2.6462,
+      "step": 621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988019235718071e-05,
+      "loss": 2.4839,
+      "step": 622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879565534766973e-05,
+      "loss": 2.4903,
+      "step": 623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878937086824162e-05,
+      "loss": 2.6098,
+      "step": 624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987830701345568e-05,
+      "loss": 2.5391,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987767531476519e-05,
+      "loss": 2.6214,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877041990856636e-05,
+      "loss": 2.6107,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876407041834216e-05,
+      "loss": 2.6282,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875770467802404e-05,
+      "loss": 2.5851,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875132268865936e-05,
+      "loss": 2.7118,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987449244512982e-05,
+      "loss": 2.6968,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873850996699325e-05,
+      "loss": 2.6441,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873207923679997e-05,
+      "loss": 2.5819,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872563226177643e-05,
+      "loss": 2.5596,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871916904298337e-05,
+      "loss": 2.6342,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871268958148416e-05,
+      "loss": 2.51,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870619387834497e-05,
+      "loss": 2.6095,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986996819346345e-05,
+      "loss": 2.6284,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869315375142423e-05,
+      "loss": 2.4907,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868660932978828e-05,
+      "loss": 2.609,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986800486708034e-05,
+      "loss": 2.6033,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986734717755491e-05,
+      "loss": 2.4642,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866687864510746e-05,
+      "loss": 2.5755,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866026928056325e-05,
+      "loss": 2.5239,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98653643683004e-05,
+      "loss": 2.5828,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986470018535198e-05,
+      "loss": 2.5413,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864034379320352e-05,
+      "loss": 2.5447,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986336695031506e-05,
+      "loss": 2.6386,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862697898445916e-05,
+      "loss": 2.6089,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986202722382301e-05,
+      "loss": 2.6009,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861354926556683e-05,
+      "loss": 2.6069,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860681006757557e-05,
+      "loss": 2.6204,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986000546453651e-05,
+      "loss": 2.6336,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98593283000047e-05,
+      "loss": 2.6709,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858649513273533e-05,
+      "loss": 2.5588,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98579691044547e-05,
+      "loss": 2.5805,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985728707366015e-05,
+      "loss": 2.6344,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856603421002104e-05,
+      "loss": 2.5997,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985591814659304e-05,
+      "loss": 2.6411,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855231250545713e-05,
+      "loss": 2.6214,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985454273297314e-05,
+      "loss": 2.6175,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985385259398861e-05,
+      "loss": 2.5189,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853160833705665e-05,
+      "loss": 2.5303,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985246745223813e-05,
+      "loss": 2.6492,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851772449700094e-05,
+      "loss": 2.5899,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851075826205903e-05,
+      "loss": 2.5505,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850377581870173e-05,
+      "loss": 2.5987,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98496777168078e-05,
+      "loss": 2.599,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848976231133923e-05,
+      "loss": 2.7146,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984827312496397e-05,
+      "loss": 2.7215,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847568398413624e-05,
+      "loss": 2.6016,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846862051598836e-05,
+      "loss": 2.5014,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984615408463582e-05,
+      "loss": 2.5624,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845444497641067e-05,
+      "loss": 2.5758,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844733290731327e-05,
+      "loss": 2.5857,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984402046402362e-05,
+      "loss": 2.6377,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843306017635226e-05,
+      "loss": 2.602,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842589951683697e-05,
+      "loss": 2.6458,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841872266286854e-05,
+      "loss": 2.5305,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984115296156278e-05,
+      "loss": 2.5919,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840432037629823e-05,
+      "loss": 2.5956,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98397094946066e-05,
+      "loss": 2.6013,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838985332611996e-05,
+      "loss": 2.5166,
+      "step": 683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983825955176516e-05,
+      "loss": 2.6213,
+      "step": 684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837532152185507e-05,
+      "loss": 2.5114,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836803133992722e-05,
+      "loss": 2.5742,
+      "step": 686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983607249730675e-05,
+      "loss": 2.585,
+      "step": 687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983534024224781e-05,
+      "loss": 2.5633,
+      "step": 688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983460636893638e-05,
+      "loss": 2.5672,
+      "step": 689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983387087749321e-05,
+      "loss": 2.5465,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983313376803931e-05,
+      "loss": 2.524,
+      "step": 691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832395040695962e-05,
+      "loss": 2.595,
+      "step": 692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831654695584716e-05,
+      "loss": 2.5589,
+      "step": 693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830912732827373e-05,
+      "loss": 2.5689,
+      "step": 694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830169152546026e-05,
+      "loss": 2.5498,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829423954863006e-05,
+      "loss": 2.6358,
+      "step": 696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828677139900934e-05,
+      "loss": 2.6181,
+      "step": 697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827928707782686e-05,
+      "loss": 2.601,
+      "step": 698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827178658631395e-05,
+      "loss": 2.5479,
+      "step": 699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826426992570482e-05,
+      "loss": 2.5983,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825673709723616e-05,
+      "loss": 2.6397,
+      "step": 701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824918810214736e-05,
+      "loss": 2.53,
+      "step": 702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824162294168056e-05,
+      "loss": 2.5657,
+      "step": 703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823404161708044e-05,
+      "loss": 2.5847,
+      "step": 704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822644412959437e-05,
+      "loss": 2.6372,
+      "step": 705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821883048047244e-05,
+      "loss": 2.645,
+      "step": 706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821120067096733e-05,
+      "loss": 2.6099,
+      "step": 707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820355470233442e-05,
+      "loss": 2.674,
+      "step": 708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819589257583176e-05,
+      "loss": 2.5748,
+      "step": 709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818821429272e-05,
+      "loss": 2.5852,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818051985426247e-05,
+      "loss": 2.5659,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817280926172522e-05,
+      "loss": 2.6002,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816508251637687e-05,
+      "loss": 2.6537,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981573396194887e-05,
+      "loss": 2.5957,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981495805723348e-05,
+      "loss": 2.5642,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981418053761917e-05,
+      "loss": 2.6176,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813401403233872e-05,
+      "loss": 2.546,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981262065420578e-05,
+      "loss": 2.5473,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811838290663353e-05,
+      "loss": 2.4785,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981105431273532e-05,
+      "loss": 2.5978,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981026872055067e-05,
+      "loss": 2.6766,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980948151423866e-05,
+      "loss": 2.6365,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808692693928815e-05,
+      "loss": 2.6312,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980790225975092e-05,
+      "loss": 2.6374,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807110211835032e-05,
+      "loss": 2.6017,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806316550311462e-05,
+      "loss": 2.6314,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805521275310803e-05,
+      "loss": 2.5461,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804724386963908e-05,
+      "loss": 2.5934,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803925885401883e-05,
+      "loss": 2.5651,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803125770756114e-05,
+      "loss": 2.6259,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802324043158247e-05,
+      "loss": 2.5858,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801520702740195e-05,
+      "loss": 2.6189,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800715749634132e-05,
+      "loss": 2.587,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799909183972505e-05,
+      "loss": 2.5802,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979910100588802e-05,
+      "loss": 2.6222,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798291215513644e-05,
+      "loss": 2.5319,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797479812982627e-05,
+      "loss": 2.6021,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796666798428462e-05,
+      "loss": 2.51,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795852171984924e-05,
+      "loss": 2.5609,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795035933786047e-05,
+      "loss": 2.637,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979421808396613e-05,
+      "loss": 2.6086,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979339862265973e-05,
+      "loss": 2.5505,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979257755000169e-05,
+      "loss": 2.5776,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791754866127096e-05,
+      "loss": 2.6597,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979093057117131e-05,
+      "loss": 2.5187,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790104665269956e-05,
+      "loss": 2.5409,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789277148558924e-05,
+      "loss": 2.6031,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978844802117437e-05,
+      "loss": 2.6525,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787617283252713e-05,
+      "loss": 2.5003,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978678493493064e-05,
+      "loss": 2.6084,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97859509763451e-05,
+      "loss": 2.5717,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785115407633307e-05,
+      "loss": 2.5812,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978427822893274e-05,
+      "loss": 2.6951,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783439440381148e-05,
+      "loss": 2.7059,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782599042116538e-05,
+      "loss": 2.5038,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781757034277184e-05,
+      "loss": 2.59,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780913417001623e-05,
+      "loss": 2.6997,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780068190428663e-05,
+      "loss": 2.5618,
+      "step": 758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779221354697374e-05,
+      "loss": 2.6852,
+      "step": 759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778372909947085e-05,
+      "loss": 2.5567,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777522856317396e-05,
+      "loss": 2.5841,
+      "step": 761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977667119394817e-05,
+      "loss": 2.5599,
+      "step": 762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775817922979542e-05,
+      "loss": 2.6886,
+      "step": 763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774963043551895e-05,
+      "loss": 2.671,
+      "step": 764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774106555805887e-05,
+      "loss": 2.5085,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773248459882444e-05,
+      "loss": 2.6058,
+      "step": 766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977238875592275e-05,
+      "loss": 2.5949,
+      "step": 767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771527444068253e-05,
+      "loss": 2.5684,
+      "step": 768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770664524460678e-05,
+      "loss": 2.577,
+      "step": 769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769799997241995e-05,
+      "loss": 2.6697,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768933862554457e-05,
+      "loss": 2.5575,
+      "step": 771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768066120540562e-05,
+      "loss": 2.6361,
+      "step": 772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767196771343095e-05,
+      "loss": 2.5493,
+      "step": 773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766325815105086e-05,
+      "loss": 2.5936,
+      "step": 774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765453251969842e-05,
+      "loss": 2.6297,
+      "step": 775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764579082080924e-05,
+      "loss": 2.6296,
+      "step": 776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763703305582172e-05,
+      "loss": 2.5213,
+      "step": 777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762825922617673e-05,
+      "loss": 2.528,
+      "step": 778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976194693333179e-05,
+      "loss": 2.5241,
+      "step": 779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976106633786915e-05,
+      "loss": 2.681,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976018413637464e-05,
+      "loss": 2.563,
+      "step": 781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759300328993406e-05,
+      "loss": 2.6855,
+      "step": 782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758414915870873e-05,
+      "loss": 2.6137,
+      "step": 783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757527897152716e-05,
+      "loss": 2.5112,
+      "step": 784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975663927298488e-05,
+      "loss": 2.5732,
+      "step": 785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755749043513582e-05,
+      "loss": 2.6192,
+      "step": 786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754857208885293e-05,
+      "loss": 2.5346,
+      "step": 787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753963769246744e-05,
+      "loss": 2.5749,
+      "step": 788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753068724744945e-05,
+      "loss": 2.6312,
+      "step": 789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975217207552715e-05,
+      "loss": 2.5813,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751273821740904e-05,
+      "loss": 2.6295,
+      "step": 791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750373963533987e-05,
+      "loss": 2.5621,
+      "step": 792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749472501054464e-05,
+      "loss": 2.5715,
+      "step": 793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748569434450657e-05,
+      "loss": 2.5688,
+      "step": 794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974766476387115e-05,
+      "loss": 2.5587,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746758489464792e-05,
+      "loss": 2.6583,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745850611380697e-05,
+      "loss": 2.5754,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974494112976824e-05,
+      "loss": 2.5822,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744030044777067e-05,
+      "loss": 2.4855,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743117356557076e-05,
+      "loss": 2.5706,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742203065258444e-05,
+      "loss": 2.6695,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741287171031593e-05,
+      "loss": 2.5764,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974036967402723e-05,
+      "loss": 2.5499,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739450574396307e-05,
+      "loss": 2.6068,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738529872290054e-05,
+      "loss": 2.5804,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737607567859952e-05,
+      "loss": 2.6075,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736683661257755e-05,
+      "loss": 2.6523,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973575815263548e-05,
+      "loss": 2.4971,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97348310421454e-05,
+      "loss": 2.6634,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733902329940058e-05,
+      "loss": 2.656,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732972016172264e-05,
+      "loss": 2.6367,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973204010099508e-05,
+      "loss": 2.5823,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973110658456184e-05,
+      "loss": 2.6439,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973017146702614e-05,
+      "loss": 2.5975,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972923474854184e-05,
+      "loss": 2.7488,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728296429263065e-05,
+      "loss": 2.6008,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727356509344194e-05,
+      "loss": 2.5181,
+      "step": 817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726414988939883e-05,
+      "loss": 2.4858,
+      "step": 818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972547186820504e-05,
+      "loss": 2.6106,
+      "step": 819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724527147294846e-05,
+      "loss": 2.5755,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723580826364734e-05,
+      "loss": 2.5539,
+      "step": 821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972263290557041e-05,
+      "loss": 2.605,
+      "step": 822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721683385067837e-05,
+      "loss": 2.5439,
+      "step": 823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720732265013248e-05,
+      "loss": 2.5862,
+      "step": 824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971977954556313e-05,
+      "loss": 2.5778,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718825226874245e-05,
+      "loss": 2.5868,
+      "step": 826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717869309103605e-05,
+      "loss": 2.5964,
+      "step": 827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971691179240849e-05,
+      "loss": 2.5391,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715952676946456e-05,
+      "loss": 2.5733,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714991962875296e-05,
+      "loss": 2.5348,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971402965035309e-05,
+      "loss": 2.4766,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713065739538168e-05,
+      "loss": 2.554,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712100230589128e-05,
+      "loss": 2.6125,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971113312366483e-05,
+      "loss": 2.4929,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710164418924394e-05,
+      "loss": 2.6454,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709194116527208e-05,
+      "loss": 2.562,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970822221663292e-05,
+      "loss": 2.536,
+      "step": 837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707248719401435e-05,
+      "loss": 2.6611,
+      "step": 838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706273624992934e-05,
+      "loss": 2.5805,
+      "step": 839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705296933567854e-05,
+      "loss": 2.61,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704318645286887e-05,
+      "loss": 2.5915,
+      "step": 841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703338760311002e-05,
+      "loss": 2.576,
+      "step": 842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702357278801425e-05,
+      "loss": 2.5664,
+      "step": 843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970137420091964e-05,
+      "loss": 2.5604,
+      "step": 844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700389526827393e-05,
+      "loss": 2.6367,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699403256686702e-05,
+      "loss": 2.5671,
+      "step": 846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698415390659844e-05,
+      "loss": 2.6338,
+      "step": 847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969742592890935e-05,
+      "loss": 2.6106,
+      "step": 848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696434871598034e-05,
+      "loss": 2.549,
+      "step": 849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695442218888946e-05,
+      "loss": 2.5198,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694447970945415e-05,
+      "loss": 2.573,
+      "step": 851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969345212793103e-05,
+      "loss": 2.5984,
+      "step": 852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692454690009643e-05,
+      "loss": 2.5223,
+      "step": 853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691455657345366e-05,
+      "loss": 2.5658,
+      "step": 854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690455030102573e-05,
+      "loss": 2.6948,
+      "step": 855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96894528084459e-05,
+      "loss": 2.5742,
+      "step": 856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688448992540255e-05,
+      "loss": 2.5018,
+      "step": 857
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968744358255079e-05,
+      "loss": 2.6518,
+      "step": 858
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686436578642938e-05,
+      "loss": 2.5682,
+      "step": 859
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685427980982382e-05,
+      "loss": 2.5601,
+      "step": 860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968441778973507e-05,
+      "loss": 2.579,
+      "step": 861
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683406005067214e-05,
+      "loss": 2.5838,
+      "step": 862
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682392627145286e-05,
+      "loss": 2.5319,
+      "step": 863
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681377656136025e-05,
+      "loss": 2.5997,
+      "step": 864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968036109220643e-05,
+      "loss": 2.6301,
+      "step": 865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967934293552376e-05,
+      "loss": 2.5403,
+      "step": 866
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967832318625553e-05,
+      "loss": 2.6326,
+      "step": 867
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967730184456953e-05,
+      "loss": 2.4785,
+      "step": 868
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967627891063381e-05,
+      "loss": 2.6399,
+      "step": 869
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675254384616664e-05,
+      "loss": 2.5577,
+      "step": 870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674228266686675e-05,
+      "loss": 2.5989,
+      "step": 871
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673200557012674e-05,
+      "loss": 2.5808,
+      "step": 872
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672171255763745e-05,
+      "loss": 2.6522,
+      "step": 873
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671140363109253e-05,
+      "loss": 2.5199,
+      "step": 874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670107879218813e-05,
+      "loss": 2.5577,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96690738042623e-05,
+      "loss": 2.6172,
+      "step": 876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966803813840986e-05,
+      "loss": 2.5631,
+      "step": 877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667000881831893e-05,
+      "loss": 2.6108,
+      "step": 878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665962034699068e-05,
+      "loss": 2.5689,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664921597182303e-05,
+      "loss": 2.5468,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663879569452792e-05,
+      "loss": 2.5321,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966283595168198e-05,
+      "loss": 2.5846,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661790744041586e-05,
+      "loss": 2.5411,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660743946703572e-05,
+      "loss": 2.52,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659695559840183e-05,
+      "loss": 2.5916,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658645583623907e-05,
+      "loss": 2.5214,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657594018227505e-05,
+      "loss": 2.5479,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656540863823993e-05,
+      "loss": 2.5243,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655486120586656e-05,
+      "loss": 2.622,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965442978868903e-05,
+      "loss": 2.5538,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965337186830492e-05,
+      "loss": 2.4794,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965231235960839e-05,
+      "loss": 2.5569,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651251262773773e-05,
+      "loss": 2.6796,
+      "step": 893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650188577975643e-05,
+      "loss": 2.5922,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649124305388856e-05,
+      "loss": 2.576,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964805844518852e-05,
+      "loss": 2.5846,
+      "step": 896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646990997550007e-05,
+      "loss": 2.5865,
+      "step": 897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645921962648952e-05,
+      "loss": 2.5379,
+      "step": 898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964485134066124e-05,
+      "loss": 2.5535,
+      "step": 899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643779131763028e-05,
+      "loss": 2.5122,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642705336130735e-05,
+      "loss": 2.6354,
+      "step": 901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641629953941032e-05,
+      "loss": 2.6521,
+      "step": 902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640552985370865e-05,
+      "loss": 2.6289,
+      "step": 903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963947443059742e-05,
+      "loss": 2.6444,
+      "step": 904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963839428979817e-05,
+      "loss": 2.4391,
+      "step": 905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637312563150826e-05,
+      "loss": 2.5065,
+      "step": 906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963622925083337e-05,
+      "loss": 2.6284,
+      "step": 907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963514435302405e-05,
+      "loss": 2.5957,
+      "step": 908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634057869901368e-05,
+      "loss": 2.6079,
+      "step": 909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963296980164408e-05,
+      "loss": 2.567,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631880148431214e-05,
+      "loss": 2.6131,
+      "step": 911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630788910442063e-05,
+      "loss": 2.4447,
+      "step": 912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629696087856163e-05,
+      "loss": 2.5936,
+      "step": 913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962860168085333e-05,
+      "loss": 2.533,
+      "step": 914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627505689613627e-05,
+      "loss": 2.5581,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962640811431738e-05,
+      "loss": 2.5671,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625308955145185e-05,
+      "loss": 2.4917,
+      "step": 917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624208212277884e-05,
+      "loss": 2.5498,
+      "step": 918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623105885896593e-05,
+      "loss": 2.5131,
+      "step": 919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962200197618268e-05,
+      "loss": 2.5825,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620896483317772e-05,
+      "loss": 2.6482,
+      "step": 921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619789407483768e-05,
+      "loss": 2.5789,
+      "step": 922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618680748862816e-05,
+      "loss": 2.5877,
+      "step": 923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617570507637333e-05,
+      "loss": 2.6931,
+      "step": 924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961645868398999e-05,
+      "loss": 2.6104,
+      "step": 925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615345278103713e-05,
+      "loss": 2.6538,
+      "step": 926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614230290161705e-05,
+      "loss": 2.5386,
+      "step": 927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613113720347416e-05,
+      "loss": 2.6135,
+      "step": 928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611995568844563e-05,
+      "loss": 2.5706,
+      "step": 929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610875835837117e-05,
+      "loss": 2.5934,
+      "step": 930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609754521509313e-05,
+      "loss": 2.5539,
+      "step": 931
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608631626045648e-05,
+      "loss": 2.49,
+      "step": 932
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607507149630875e-05,
+      "loss": 2.5657,
+      "step": 933
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960638109245001e-05,
+      "loss": 2.5442,
+      "step": 934
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605253454688328e-05,
+      "loss": 2.5652,
+      "step": 935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604124236531365e-05,
+      "loss": 2.5804,
+      "step": 936
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960299343816492e-05,
+      "loss": 2.6134,
+      "step": 937
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601861059775037e-05,
+      "loss": 2.6996,
+      "step": 938
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600727101548043e-05,
+      "loss": 2.5603,
+      "step": 939
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599591563670506e-05,
+      "loss": 2.6168,
+      "step": 940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598454446329266e-05,
+      "loss": 2.5553,
+      "step": 941
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597315749711413e-05,
+      "loss": 2.6397,
+      "step": 942
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596175474004305e-05,
+      "loss": 2.6087,
+      "step": 943
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959503361939556e-05,
+      "loss": 2.5642,
+      "step": 944
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593890186073045e-05,
+      "loss": 2.4873,
+      "step": 945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592745174224902e-05,
+      "loss": 2.6235,
+      "step": 946
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591598584039515e-05,
+      "loss": 2.4841,
+      "step": 947
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590450415705547e-05,
+      "loss": 2.527,
+      "step": 948
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589300669411906e-05,
+      "loss": 2.5932,
+      "step": 949
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588149345347767e-05,
+      "loss": 2.5935,
+      "step": 950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586996443702563e-05,
+      "loss": 2.5894,
+      "step": 951
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585841964665983e-05,
+      "loss": 2.6401,
+      "step": 952
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584685908427984e-05,
+      "loss": 2.5209,
+      "step": 953
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958352827517877e-05,
+      "loss": 2.6291,
+      "step": 954
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958236906510882e-05,
+      "loss": 2.5911,
+      "step": 955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581208278408857e-05,
+      "loss": 2.5893,
+      "step": 956
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580045915269873e-05,
+      "loss": 2.5098,
+      "step": 957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578881975883116e-05,
+      "loss": 2.5838,
+      "step": 958
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577716460440098e-05,
+      "loss": 2.5672,
+      "step": 959
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957654936913258e-05,
+      "loss": 2.6396,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957538070215259e-05,
+      "loss": 2.6006,
+      "step": 961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957421045969242e-05,
+      "loss": 2.5814,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957303864194461e-05,
+      "loss": 2.6033,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571865249101964e-05,
+      "loss": 2.5949,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957069028135755e-05,
+      "loss": 2.5618,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569513738904687e-05,
+      "loss": 2.5481,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568335621936956e-05,
+      "loss": 2.5057,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95671559306482e-05,
+      "loss": 2.544,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565974665232522e-05,
+      "loss": 2.63,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956479182588427e-05,
+      "loss": 2.5723,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563607412798077e-05,
+      "loss": 2.6223,
+      "step": 971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562421426168807e-05,
+      "loss": 2.5748,
+      "step": 972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95612338661916e-05,
+      "loss": 2.649,
+      "step": 973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560044733061855e-05,
+      "loss": 2.6315,
+      "step": 974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955885402697522e-05,
+      "loss": 2.5313,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557661748127608e-05,
+      "loss": 2.5745,
+      "step": 976
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556467896715192e-05,
+      "loss": 2.5549,
+      "step": 977
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555272472934397e-05,
+      "loss": 2.5965,
+      "step": 978
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955407547698192e-05,
+      "loss": 2.5118,
+      "step": 979
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95528769090547e-05,
+      "loss": 2.5888,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551676769349945e-05,
+      "loss": 2.5237,
+      "step": 981
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550475058065122e-05,
+      "loss": 2.6038,
+      "step": 982
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954927177539795e-05,
+      "loss": 2.6813,
+      "step": 983
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548066921546412e-05,
+      "loss": 2.5333,
+      "step": 984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954686049670875e-05,
+      "loss": 2.699,
+      "step": 985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545652501083462e-05,
+      "loss": 2.5693,
+      "step": 986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544442934869302e-05,
+      "loss": 2.501,
+      "step": 987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954323179826529e-05,
+      "loss": 2.5644,
+      "step": 988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954201909147069e-05,
+      "loss": 2.5565,
+      "step": 989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954080481468505e-05,
+      "loss": 2.5503,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539588968108144e-05,
+      "loss": 2.5126,
+      "step": 991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953837155194003e-05,
+      "loss": 2.5525,
+      "step": 992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537152566381013e-05,
+      "loss": 2.5677,
+      "step": 993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953593201163166e-05,
+      "loss": 2.6316,
+      "step": 994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534709887892793e-05,
+      "loss": 2.5399,
+      "step": 995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953348619536549e-05,
+      "loss": 2.5452,
+      "step": 996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532260934251093e-05,
+      "loss": 2.6356,
+      "step": 997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531034104751202e-05,
+      "loss": 2.5441,
+      "step": 998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952980570706767e-05,
+      "loss": 2.587,
+      "step": 999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528575741402612e-05,
+      "loss": 2.5909,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527344207958395e-05,
+      "loss": 2.6167,
+      "step": 1001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526111106937658e-05,
+      "loss": 2.6158,
+      "step": 1002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524876438543286e-05,
+      "loss": 2.5374,
+      "step": 1003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952364020297842e-05,
+      "loss": 2.5689,
+      "step": 1004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522402400446464e-05,
+      "loss": 2.5266,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952116303115108e-05,
+      "loss": 2.6321,
+      "step": 1006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951992209529619e-05,
+      "loss": 2.6087,
+      "step": 1007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518679593085967e-05,
+      "loss": 2.4421,
+      "step": 1008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951743552472485e-05,
+      "loss": 2.6096,
+      "step": 1009
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516189890417525e-05,
+      "loss": 2.5968,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514942690368945e-05,
+      "loss": 2.5632,
+      "step": 1011
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513693924784317e-05,
+      "loss": 2.4616,
+      "step": 1012
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512443593869105e-05,
+      "loss": 2.5985,
+      "step": 1013
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511191697829038e-05,
+      "loss": 2.5181,
+      "step": 1014
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509938236870085e-05,
+      "loss": 2.5512,
+      "step": 1015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508683211198492e-05,
+      "loss": 2.5347,
+      "step": 1016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507426621020752e-05,
+      "loss": 2.4408,
+      "step": 1017
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506168466543615e-05,
+      "loss": 2.4867,
+      "step": 1018
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504908747974093e-05,
+      "loss": 2.6244,
+      "step": 1019
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503647465519454e-05,
+      "loss": 2.5626,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950238461938722e-05,
+      "loss": 2.6198,
+      "step": 1021
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501120209785172e-05,
+      "loss": 2.5647,
+      "step": 1022
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499854236921353e-05,
+      "loss": 2.5954,
+      "step": 1023
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498586701004055e-05,
+      "loss": 2.4997,
+      "step": 1024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497317602241835e-05,
+      "loss": 2.5805,
+      "step": 1025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94960469408435e-05,
+      "loss": 2.585,
+      "step": 1026
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949477471701812e-05,
+      "loss": 2.5366,
+      "step": 1027
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493500930975023e-05,
+      "loss": 2.6171,
+      "step": 1028
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949222558292378e-05,
+      "loss": 2.5974,
+      "step": 1029
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490948673074238e-05,
+      "loss": 2.6011,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489670201636493e-05,
+      "loss": 2.5468,
+      "step": 1031
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488390168820897e-05,
+      "loss": 2.6366,
+      "step": 1032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487108574838056e-05,
+      "loss": 2.5513,
+      "step": 1033
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485825419898837e-05,
+      "loss": 2.5469,
+      "step": 1034
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948454070421437e-05,
+      "loss": 2.573,
+      "step": 1035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483254427996025e-05,
+      "loss": 2.6334,
+      "step": 1036
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481966591455448e-05,
+      "loss": 2.5998,
+      "step": 1037
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948067719480453e-05,
+      "loss": 2.4748,
+      "step": 1038
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947938623825541e-05,
+      "loss": 2.6205,
+      "step": 1039
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947809372202051e-05,
+      "loss": 2.5581,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947679964631249e-05,
+      "loss": 2.5874,
+      "step": 1041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475504011344263e-05,
+      "loss": 2.4705,
+      "step": 1042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947420681732901e-05,
+      "loss": 2.549,
+      "step": 1043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472908064480166e-05,
+      "loss": 2.6047,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947160775301142e-05,
+      "loss": 2.5162,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470305883136716e-05,
+      "loss": 2.5485,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946900245507026e-05,
+      "loss": 2.5001,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946769746902651e-05,
+      "loss": 2.5699,
+      "step": 1048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466390925220173e-05,
+      "loss": 2.5783,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465082823866232e-05,
+      "loss": 2.5887,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463773165179906e-05,
+      "loss": 2.6318,
+      "step": 1051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462461949376684e-05,
+      "loss": 2.6383,
+      "step": 1052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946114917667231e-05,
+      "loss": 2.526,
+      "step": 1053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945983484728277e-05,
+      "loss": 2.5876,
+      "step": 1054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458518961424323e-05,
+      "loss": 2.5675,
+      "step": 1055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457201519313478e-05,
+      "loss": 2.5737,
+      "step": 1056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455882521166998e-05,
+      "loss": 2.5556,
+      "step": 1057
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454561967201903e-05,
+      "loss": 2.4793,
+      "step": 1058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945323985763547e-05,
+      "loss": 2.5049,
+      "step": 1059
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451916192685233e-05,
+      "loss": 2.4743,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945059097256898e-05,
+      "loss": 2.6194,
+      "step": 1061
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449264197504754e-05,
+      "loss": 2.5899,
+      "step": 1062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447935867710858e-05,
+      "loss": 2.5968,
+      "step": 1063
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446605983405843e-05,
+      "loss": 2.5265,
+      "step": 1064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445274544808526e-05,
+      "loss": 2.6059,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944394155213797e-05,
+      "loss": 2.5985,
+      "step": 1066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442607005613506e-05,
+      "loss": 2.5131,
+      "step": 1067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441270905454705e-05,
+      "loss": 2.5979,
+      "step": 1068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439933251881405e-05,
+      "loss": 2.5702,
+      "step": 1069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438594045113694e-05,
+      "loss": 2.5589,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437253285371918e-05,
+      "loss": 2.5975,
+      "step": 1071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943591097287668e-05,
+      "loss": 2.6235,
+      "step": 1072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434567107848838e-05,
+      "loss": 2.689,
+      "step": 1073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94332216905095e-05,
+      "loss": 2.6574,
+      "step": 1074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431874721080034e-05,
+      "loss": 2.5266,
+      "step": 1075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943052619978207e-05,
+      "loss": 2.6365,
+      "step": 1076
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429176126837476e-05,
+      "loss": 2.562,
+      "step": 1077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427824502468394e-05,
+      "loss": 2.6135,
+      "step": 1078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426471326897206e-05,
+      "loss": 2.6058,
+      "step": 1079
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942511660034656e-05,
+      "loss": 2.6058,
+      "step": 1080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942376032303935e-05,
+      "loss": 2.6484,
+      "step": 1081
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422402495198742e-05,
+      "loss": 2.6185,
+      "step": 1082
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421043117048132e-05,
+      "loss": 2.5328,
+      "step": 1083
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419682188811197e-05,
+      "loss": 2.536,
+      "step": 1084
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418319710711846e-05,
+      "loss": 2.5051,
+      "step": 1085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941695568297426e-05,
+      "loss": 2.5003,
+      "step": 1086
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415590105822862e-05,
+      "loss": 2.591,
+      "step": 1087
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414222979482345e-05,
+      "loss": 2.6048,
+      "step": 1088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412854304177645e-05,
+      "loss": 2.5926,
+      "step": 1089
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411484080133953e-05,
+      "loss": 2.6065,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410112307576723e-05,
+      "loss": 2.4895,
+      "step": 1091
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408738986731655e-05,
+      "loss": 2.5463,
+      "step": 1092
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407364117824707e-05,
+      "loss": 2.621,
+      "step": 1093
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405987701082097e-05,
+      "loss": 2.5479,
+      "step": 1094
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404609736730293e-05,
+      "loss": 2.5651,
+      "step": 1095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403230224996013e-05,
+      "loss": 2.6485,
+      "step": 1096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401849166106236e-05,
+      "loss": 2.4694,
+      "step": 1097
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400466560288197e-05,
+      "loss": 2.6161,
+      "step": 1098
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399082407769375e-05,
+      "loss": 2.5212,
+      "step": 1099
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397696708777517e-05,
+      "loss": 2.4881,
+      "step": 1100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939630946354062e-05,
+      "loss": 2.5687,
+      "step": 1101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394920672286924e-05,
+      "loss": 2.5407,
+      "step": 1102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393530335244946e-05,
+      "loss": 2.6446,
+      "step": 1103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392138452643434e-05,
+      "loss": 2.5776,
+      "step": 1104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390745024711407e-05,
+      "loss": 2.6621,
+      "step": 1105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389350051678125e-05,
+      "loss": 2.5848,
+      "step": 1106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387953533773115e-05,
+      "loss": 2.5681,
+      "step": 1107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386555471226152e-05,
+      "loss": 2.5401,
+      "step": 1108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938515586426726e-05,
+      "loss": 2.6286,
+      "step": 1109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938375471312673e-05,
+      "loss": 2.5794,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382352018035098e-05,
+      "loss": 2.5349,
+      "step": 1111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380947779223153e-05,
+      "loss": 2.5974,
+      "step": 1112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937954199692194e-05,
+      "loss": 2.5544,
+      "step": 1113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378134671362762e-05,
+      "loss": 2.6236,
+      "step": 1114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376725802777172e-05,
+      "loss": 2.6025,
+      "step": 1115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375315391396976e-05,
+      "loss": 2.4964,
+      "step": 1116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373903437454237e-05,
+      "loss": 2.5773,
+      "step": 1117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372489941181264e-05,
+      "loss": 2.569,
+      "step": 1118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371074902810637e-05,
+      "loss": 2.5414,
+      "step": 1119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369658322575168e-05,
+      "loss": 2.5174,
+      "step": 1120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368240200707938e-05,
+      "loss": 2.647,
+      "step": 1121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366820537442277e-05,
+      "loss": 2.559,
+      "step": 1122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936539933301177e-05,
+      "loss": 2.553,
+      "step": 1123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936397658765025e-05,
+      "loss": 2.7877,
+      "step": 1124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936255230159181e-05,
+      "loss": 2.5574,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361126475070793e-05,
+      "loss": 2.602,
+      "step": 1126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93596991083218e-05,
+      "loss": 2.488,
+      "step": 1127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358270201579678e-05,
+      "loss": 2.5972,
+      "step": 1128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356839755079532e-05,
+      "loss": 2.478,
+      "step": 1129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355407769056724e-05,
+      "loss": 2.5949,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935397424374686e-05,
+      "loss": 2.5984,
+      "step": 1131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352539179385808e-05,
+      "loss": 2.6331,
+      "step": 1132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935110257620968e-05,
+      "loss": 2.6007,
+      "step": 1133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349664434454854e-05,
+      "loss": 2.5348,
+      "step": 1134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934822475435795e-05,
+      "loss": 2.5678,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346783536155843e-05,
+      "loss": 2.5812,
+      "step": 1136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345340780085667e-05,
+      "loss": 2.5685,
+      "step": 1137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343896486384803e-05,
+      "loss": 2.6485,
+      "step": 1138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342450655290888e-05,
+      "loss": 2.5493,
+      "step": 1139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934100328704181e-05,
+      "loss": 2.5771,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933955438187571e-05,
+      "loss": 2.6831,
+      "step": 1141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338103940030987e-05,
+      "loss": 2.6,
+      "step": 1142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336651961746287e-05,
+      "loss": 2.5193,
+      "step": 1143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335198447260507e-05,
+      "loss": 2.6057,
+      "step": 1144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333743396812806e-05,
+      "loss": 2.5512,
+      "step": 1145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332286810642585e-05,
+      "loss": 2.6034,
+      "step": 1146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330828688989506e-05,
+      "loss": 2.5742,
+      "step": 1147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329369032093475e-05,
+      "loss": 2.5641,
+      "step": 1148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932790784019466e-05,
+      "loss": 2.605,
+      "step": 1149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326445113533482e-05,
+      "loss": 2.5197,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324980852350604e-05,
+      "loss": 2.6699,
+      "step": 1151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323515056886946e-05,
+      "loss": 2.5306,
+      "step": 1152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322047727383683e-05,
+      "loss": 2.5497,
+      "step": 1153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320578864082246e-05,
+      "loss": 2.5467,
+      "step": 1154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319108467224307e-05,
+      "loss": 2.5322,
+      "step": 1155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317636537051806e-05,
+      "loss": 2.6194,
+      "step": 1156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316163073806912e-05,
+      "loss": 2.6018,
+      "step": 1157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314688077732075e-05,
+      "loss": 2.6302,
+      "step": 1158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313211549069975e-05,
+      "loss": 2.5711,
+      "step": 1159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311733488063552e-05,
+      "loss": 2.5441,
+      "step": 1160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310253894956003e-05,
+      "loss": 2.5198,
+      "step": 1161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308772769990768e-05,
+      "loss": 2.5974,
+      "step": 1162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307290113411542e-05,
+      "loss": 2.6024,
+      "step": 1163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305805925462273e-05,
+      "loss": 2.5801,
+      "step": 1164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304320206387166e-05,
+      "loss": 2.4826,
+      "step": 1165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930283295643067e-05,
+      "loss": 2.5503,
+      "step": 1166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301344175837485e-05,
+      "loss": 2.4782,
+      "step": 1167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299853864852575e-05,
+      "loss": 2.503,
+      "step": 1168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929836202372114e-05,
+      "loss": 2.5125,
+      "step": 1169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296868652688645e-05,
+      "loss": 2.6425,
+      "step": 1170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295373752000795e-05,
+      "loss": 2.6625,
+      "step": 1171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293877321903557e-05,
+      "loss": 2.5765,
+      "step": 1172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292379362643147e-05,
+      "loss": 2.5255,
+      "step": 1173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290879874466028e-05,
+      "loss": 2.6087,
+      "step": 1174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289378857618916e-05,
+      "loss": 2.5584,
+      "step": 1175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928787631234878e-05,
+      "loss": 2.6089,
+      "step": 1176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286372238902843e-05,
+      "loss": 2.5898,
+      "step": 1177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928486663752858e-05,
+      "loss": 2.5598,
+      "step": 1178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283359508473706e-05,
+      "loss": 2.5252,
+      "step": 1179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281850851986205e-05,
+      "loss": 2.6507,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280340668314293e-05,
+      "loss": 2.63,
+      "step": 1181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278828957706454e-05,
+      "loss": 2.5456,
+      "step": 1182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277315720411415e-05,
+      "loss": 2.5802,
+      "step": 1183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275800956678156e-05,
+      "loss": 2.6279,
+      "step": 1184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274284666755905e-05,
+      "loss": 2.5767,
+      "step": 1185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927276685089415e-05,
+      "loss": 2.639,
+      "step": 1186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927124750934262e-05,
+      "loss": 2.5289,
+      "step": 1187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269726642351302e-05,
+      "loss": 2.6301,
+      "step": 1188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268204250170423e-05,
+      "loss": 2.5712,
+      "step": 1189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266680333050477e-05,
+      "loss": 2.4233,
+      "step": 1190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265154891242203e-05,
+      "loss": 2.5746,
+      "step": 1191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263627924996577e-05,
+      "loss": 2.5198,
+      "step": 1192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262099434564847e-05,
+      "loss": 2.5337,
+      "step": 1193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92605694201985e-05,
+      "loss": 2.5452,
+      "step": 1194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259037882149277e-05,
+      "loss": 2.6437,
+      "step": 1195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925750482066917e-05,
+      "loss": 2.6162,
+      "step": 1196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255970236010418e-05,
+      "loss": 2.4471,
+      "step": 1197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925443412842551e-05,
+      "loss": 2.5035,
+      "step": 1198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252896498167194e-05,
+      "loss": 2.5891,
+      "step": 1199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251357345488465e-05,
+      "loss": 2.5052,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249816670642558e-05,
+      "loss": 2.4831,
+      "step": 1201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248274473882976e-05,
+      "loss": 2.5599,
+      "step": 1202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246730755463456e-05,
+      "loss": 2.634,
+      "step": 1203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245185515638e-05,
+      "loss": 2.5475,
+      "step": 1204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243638754660847e-05,
+      "loss": 2.5681,
+      "step": 1205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92420904727865e-05,
+      "loss": 2.5683,
+      "step": 1206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240540670269693e-05,
+      "loss": 2.624,
+      "step": 1207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238989347365437e-05,
+      "loss": 2.4826,
+      "step": 1208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237436504328965e-05,
+      "loss": 2.601,
+      "step": 1209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235882141415777e-05,
+      "loss": 2.5626,
+      "step": 1210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234326258881625e-05,
+      "loss": 2.5306,
+      "step": 1211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232768856982497e-05,
+      "loss": 2.5136,
+      "step": 1212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231209935974647e-05,
+      "loss": 2.6703,
+      "step": 1213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229649496114563e-05,
+      "loss": 2.5141,
+      "step": 1214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228087537659e-05,
+      "loss": 2.5538,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226524060864947e-05,
+      "loss": 2.493,
+      "step": 1216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224959065989654e-05,
+      "loss": 2.6108,
+      "step": 1217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223392553290613e-05,
+      "loss": 2.638,
+      "step": 1218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922182452302557e-05,
+      "loss": 2.5999,
+      "step": 1219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220254975452524e-05,
+      "loss": 2.6355,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921868391082972e-05,
+      "loss": 2.628,
+      "step": 1221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217111329415643e-05,
+      "loss": 2.5696,
+      "step": 1222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921553723146905e-05,
+      "loss": 2.5786,
+      "step": 1223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921396161724892e-05,
+      "loss": 2.5985,
+      "step": 1224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212384487014507e-05,
+      "loss": 2.6415,
+      "step": 1225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92108058410253e-05,
+      "loss": 2.5176,
+      "step": 1226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209225679541043e-05,
+      "loss": 2.6548,
+      "step": 1227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920764400282172e-05,
+      "loss": 2.5749,
+      "step": 1228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920606081112758e-05,
+      "loss": 2.4851,
+      "step": 1229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204476104719103e-05,
+      "loss": 2.4546,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202889883857036e-05,
+      "loss": 2.5541,
+      "step": 1231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201302148802365e-05,
+      "loss": 2.557,
+      "step": 1232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199712899816325e-05,
+      "loss": 2.5293,
+      "step": 1233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198122137160403e-05,
+      "loss": 2.6744,
+      "step": 1234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196529861096338e-05,
+      "loss": 2.5482,
+      "step": 1235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194936071886107e-05,
+      "loss": 2.6162,
+      "step": 1236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919334076979195e-05,
+      "loss": 2.5092,
+      "step": 1237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191743955076345e-05,
+      "loss": 2.5172,
+      "step": 1238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190145628002025e-05,
+      "loss": 2.5271,
+      "step": 1239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188545788831973e-05,
+      "loss": 2.5691,
+      "step": 1240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918694443782941e-05,
+      "loss": 2.5218,
+      "step": 1241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185341575257818e-05,
+      "loss": 2.5615,
+      "step": 1242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183737201380923e-05,
+      "loss": 2.6694,
+      "step": 1243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182131316462702e-05,
+      "loss": 2.5822,
+      "step": 1244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180523920767372e-05,
+      "loss": 2.5101,
+      "step": 1245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178915014559414e-05,
+      "loss": 2.5564,
+      "step": 1246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917730459810354e-05,
+      "loss": 2.5951,
+      "step": 1247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175692671664725e-05,
+      "loss": 2.5906,
+      "step": 1248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174079235508183e-05,
+      "loss": 2.5526,
+      "step": 1249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917246428989938e-05,
+      "loss": 2.5786,
+      "step": 1250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170847835104035e-05,
+      "loss": 2.6078,
+      "step": 1251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91692298713881e-05,
+      "loss": 2.6367,
+      "step": 1252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167610399017797e-05,
+      "loss": 2.6249,
+      "step": 1253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916598941825958e-05,
+      "loss": 2.5399,
+      "step": 1254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164366929380155e-05,
+      "loss": 2.5356,
+      "step": 1255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916274293264648e-05,
+      "loss": 2.4526,
+      "step": 1256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916111742832576e-05,
+      "loss": 2.5495,
+      "step": 1257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159490416685442e-05,
+      "loss": 2.5221,
+      "step": 1258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157861897993226e-05,
+      "loss": 2.5552,
+      "step": 1259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156231872517065e-05,
+      "loss": 2.4688,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154600340525146e-05,
+      "loss": 2.5348,
+      "step": 1261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152967302285917e-05,
+      "loss": 2.5278,
+      "step": 1262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915133275806807e-05,
+      "loss": 2.5858,
+      "step": 1263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149696708140543e-05,
+      "loss": 2.6652,
+      "step": 1264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914805915277252e-05,
+      "loss": 2.5774,
+      "step": 1265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914642009223344e-05,
+      "loss": 2.507,
+      "step": 1266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914477952679298e-05,
+      "loss": 2.4684,
+      "step": 1267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143137456721074e-05,
+      "loss": 2.5874,
+      "step": 1268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141493882287896e-05,
+      "loss": 2.6084,
+      "step": 1269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139848803763872e-05,
+      "loss": 2.544,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138202221419676e-05,
+      "loss": 2.641,
+      "step": 1271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913655413552622e-05,
+      "loss": 2.6342,
+      "step": 1272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913490454635468e-05,
+      "loss": 2.5659,
+      "step": 1273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133253454176464e-05,
+      "loss": 2.4664,
+      "step": 1274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913160085926324e-05,
+      "loss": 2.6432,
+      "step": 1275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129946761886912e-05,
+      "loss": 2.6559,
+      "step": 1276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128291162319637e-05,
+      "loss": 2.6107,
+      "step": 1277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912663406083382e-05,
+      "loss": 2.5506,
+      "step": 1278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124975457702106e-05,
+      "loss": 2.5225,
+      "step": 1279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123315353197396e-05,
+      "loss": 2.6032,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9121653747592836e-05,
+      "loss": 2.5542,
+      "step": 1281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119990641161816e-05,
+      "loss": 2.4916,
+      "step": 1282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118326034177973e-05,
+      "loss": 2.5373,
+      "step": 1283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9116659926915192e-05,
+      "loss": 2.5166,
+      "step": 1284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911499231964761e-05,
+      "loss": 2.7556,
+      "step": 1285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91133232126496e-05,
+      "loss": 2.5349,
+      "step": 1286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911165260619579e-05,
+      "loss": 2.555,
+      "step": 1287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109980500561053e-05,
+      "loss": 2.4171,
+      "step": 1288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108306896020506e-05,
+      "loss": 2.5688,
+      "step": 1289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9106631792849516e-05,
+      "loss": 2.5165,
+      "step": 1290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104955191323697e-05,
+      "loss": 2.5494,
+      "step": 1291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103277091718904e-05,
+      "loss": 2.5894,
+      "step": 1292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101597494311245e-05,
+      "loss": 2.5239,
+      "step": 1293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099916399377072e-05,
+      "loss": 2.5476,
+      "step": 1294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909823380719298e-05,
+      "loss": 2.5786,
+      "step": 1295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096549718035816e-05,
+      "loss": 2.5192,
+      "step": 1296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909486413218267e-05,
+      "loss": 2.5726,
+      "step": 1297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093177049910877e-05,
+      "loss": 2.5986,
+      "step": 1298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091488471498026e-05,
+      "loss": 2.6039,
+      "step": 1299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089798397221938e-05,
+      "loss": 2.6191,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9088106827360696e-05,
+      "loss": 2.5655,
+      "step": 1301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086413762192616e-05,
+      "loss": 2.6362,
+      "step": 1302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908471920199627e-05,
+      "loss": 2.5475,
+      "step": 1303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908302314705047e-05,
+      "loss": 2.532,
+      "step": 1304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081325597634275e-05,
+      "loss": 2.5534,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907962655402699e-05,
+      "loss": 2.5944,
+      "step": 1306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077926016508168e-05,
+      "loss": 2.6725,
+      "step": 1307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90762239853576e-05,
+      "loss": 2.5662,
+      "step": 1308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074520460855336e-05,
+      "loss": 2.5713,
+      "step": 1309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072815443281665e-05,
+      "loss": 2.5459,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071108932917114e-05,
+      "loss": 2.5792,
+      "step": 1311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906940093004247e-05,
+      "loss": 2.5652,
+      "step": 1312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067691434938756e-05,
+      "loss": 2.5493,
+      "step": 1313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906598044788724e-05,
+      "loss": 2.6586,
+      "step": 1314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064267969169443e-05,
+      "loss": 2.5121,
+      "step": 1315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062553999067122e-05,
+      "loss": 2.5406,
+      "step": 1316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906083853786229e-05,
+      "loss": 2.567,
+      "step": 1317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905912158583719e-05,
+      "loss": 2.5525,
+      "step": 1318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905740314327433e-05,
+      "loss": 2.5876,
+      "step": 1319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055683210456447e-05,
+      "loss": 2.4693,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053961787666533e-05,
+      "loss": 2.6372,
+      "step": 1321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052238875187822e-05,
+      "loss": 2.6054,
+      "step": 1322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050514473303784e-05,
+      "loss": 2.557,
+      "step": 1323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904878858229815e-05,
+      "loss": 2.4972,
+      "step": 1324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047061202454887e-05,
+      "loss": 2.5168,
+      "step": 1325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045332334058213e-05,
+      "loss": 2.6313,
+      "step": 1326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904360197739258e-05,
+      "loss": 2.5658,
+      "step": 1327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041870132742693e-05,
+      "loss": 2.5246,
+      "step": 1328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90401368003935e-05,
+      "loss": 2.5811,
+      "step": 1329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038401980630192e-05,
+      "loss": 2.6304,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036665673738213e-05,
+      "loss": 2.6467,
+      "step": 1331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034927880003237e-05,
+      "loss": 2.5559,
+      "step": 1332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90331885997112e-05,
+      "loss": 2.6464,
+      "step": 1333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031447833148265e-05,
+      "loss": 2.5915,
+      "step": 1334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029705580600853e-05,
+      "loss": 2.5414,
+      "step": 1335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027961842355626e-05,
+      "loss": 2.5991,
+      "step": 1336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9026216618699487e-05,
+      "loss": 2.5384,
+      "step": 1337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024469909919583e-05,
+      "loss": 2.5575,
+      "step": 1338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902272171630331e-05,
+      "loss": 2.578,
+      "step": 1339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902097203813831e-05,
+      "loss": 2.5364,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019220875712453e-05,
+      "loss": 2.4286,
+      "step": 1341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901746822931388e-05,
+      "loss": 2.5235,
+      "step": 1342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015714099230957e-05,
+      "loss": 2.5038,
+      "step": 1343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013958485752294e-05,
+      "loss": 2.5003,
+      "step": 1344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012201389166757e-05,
+      "loss": 2.5714,
+      "step": 1345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901044280976344e-05,
+      "loss": 2.6284,
+      "step": 1346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9008682747831698e-05,
+      "loss": 2.4431,
+      "step": 1347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900692120366112e-05,
+      "loss": 2.5711,
+      "step": 1348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005158177541542e-05,
+      "loss": 2.5665,
+      "step": 1349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900339366976304e-05,
+      "loss": 2.5694,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001627680615932e-05,
+      "loss": 2.5757,
+      "step": 1351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899986021039079e-05,
+      "loss": 2.6008,
+      "step": 1352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998091259378428e-05,
+      "loss": 2.6027,
+      "step": 1353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996320827869895e-05,
+      "loss": 2.6022,
+      "step": 1354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899454891615648e-05,
+      "loss": 2.5594,
+      "step": 1355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992775524529734e-05,
+      "loss": 2.5322,
+      "step": 1356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899100065328144e-05,
+      "loss": 2.4806,
+      "step": 1357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989224302703622e-05,
+      "loss": 2.6533,
+      "step": 1358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898744647308855e-05,
+      "loss": 2.5619,
+      "step": 1359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898566716472874e-05,
+      "loss": 2.5368,
+      "step": 1360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983886377916954e-05,
+      "loss": 2.5354,
+      "step": 1361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898210411294619e-05,
+      "loss": 2.5611,
+      "step": 1362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980320370109688e-05,
+      "loss": 2.5305,
+      "step": 1363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978535149700938e-05,
+      "loss": 2.5653,
+      "step": 1364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976748452013673e-05,
+      "loss": 2.6052,
+      "step": 1365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974960277341858e-05,
+      "loss": 2.5461,
+      "step": 1366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897317062597972e-05,
+      "loss": 2.568,
+      "step": 1367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971379498221712e-05,
+      "loss": 2.5253,
+      "step": 1368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969586894362537e-05,
+      "loss": 2.5246,
+      "step": 1369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896779281469714e-05,
+      "loss": 2.5254,
+      "step": 1370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965997259520713e-05,
+      "loss": 2.6137,
+      "step": 1371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964200229128677e-05,
+      "loss": 2.4959,
+      "step": 1372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896240172381672e-05,
+      "loss": 2.6158,
+      "step": 1373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960601743880742e-05,
+      "loss": 2.6308,
+      "step": 1374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958800289616915e-05,
+      "loss": 2.6263,
+      "step": 1375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956997361321633e-05,
+      "loss": 2.5495,
+      "step": 1376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955192959291544e-05,
+      "loss": 2.6187,
+      "step": 1377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895338708382353e-05,
+      "loss": 2.5749,
+      "step": 1378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951579735214724e-05,
+      "loss": 2.6065,
+      "step": 1379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949770913762495e-05,
+      "loss": 2.4697,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8947960619764456e-05,
+      "loss": 2.5604,
+      "step": 1381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946148853518464e-05,
+      "loss": 2.54,
+      "step": 1382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944335615322616e-05,
+      "loss": 2.566,
+      "step": 1383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894252090547525e-05,
+      "loss": 2.5091,
+      "step": 1384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940704724274957e-05,
+      "loss": 2.5221,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893888707202055e-05,
+      "loss": 2.5026,
+      "step": 1386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937067949011105e-05,
+      "loss": 2.5711,
+      "step": 1387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893524735554592e-05,
+      "loss": 2.5486,
+      "step": 1388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933425291924554e-05,
+      "loss": 2.5698,
+      "step": 1389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89316017584468e-05,
+      "loss": 2.5414,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929776755412684e-05,
+      "loss": 2.4716,
+      "step": 1391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927950283122493e-05,
+      "loss": 2.5914,
+      "step": 1392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926122341876736e-05,
+      "loss": 2.5631,
+      "step": 1393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924292931976175e-05,
+      "loss": 2.635,
+      "step": 1394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8922462053721808e-05,
+      "loss": 2.5654,
+      "step": 1395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920629707414883e-05,
+      "loss": 2.566,
+      "step": 1396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891879589335688e-05,
+      "loss": 2.6158,
+      "step": 1397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891696061184953e-05,
+      "loss": 2.489,
+      "step": 1398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915123863194794e-05,
+      "loss": 2.6004,
+      "step": 1399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891328564769488e-05,
+      "loss": 2.5624,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891144596565224e-05,
+      "loss": 2.5706,
+      "step": 1401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909604817369573e-05,
+      "loss": 2.5309,
+      "step": 1402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89077622031498e-05,
+      "loss": 2.5247,
+      "step": 1403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905918123296095e-05,
+      "loss": 2.4964,
+      "step": 1404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890407257811188e-05,
+      "loss": 2.5515,
+      "step": 1405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902225567900807e-05,
+      "loss": 2.5907,
+      "step": 1406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900377092966772e-05,
+      "loss": 2.6165,
+      "step": 1407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898527153613914e-05,
+      "loss": 2.6083,
+      "step": 1408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889667575014661e-05,
+      "loss": 2.5912,
+      "step": 1409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894822882869482e-05,
+      "loss": 2.6209,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892968552087394e-05,
+      "loss": 2.6256,
+      "step": 1411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891112758105437e-05,
+      "loss": 2.6051,
+      "step": 1412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889255501228964e-05,
+      "loss": 2.5706,
+      "step": 1413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888739678176355e-05,
+      "loss": 2.5465,
+      "step": 1414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885536600015022e-05,
+      "loss": 2.5643,
+      "step": 1415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883674956289443e-05,
+      "loss": 2.559,
+      "step": 1416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888181185089312e-05,
+      "loss": 2.5733,
+      "step": 1417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887994728413259e-05,
+      "loss": 2.599,
+      "step": 1418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887808125631465e-05,
+      "loss": 2.6373,
+      "step": 1419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887621376774632e-05,
+      "loss": 2.6179,
+      "step": 1420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874344818734868e-05,
+      "loss": 2.6532,
+      "step": 1421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872474409587794e-05,
+      "loss": 2.5702,
+      "step": 1422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870602540612853e-05,
+      "loss": 2.6202,
+      "step": 1423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886872921211803e-05,
+      "loss": 2.4582,
+      "step": 1424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886685442441155e-05,
+      "loss": 2.5201,
+      "step": 1425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886497817780188e-05,
+      "loss": 2.5622,
+      "step": 1426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886310047259773e-05,
+      "loss": 2.5506,
+      "step": 1427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8861221309108048e-05,
+      "loss": 2.5826,
+      "step": 1428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859340687642014e-05,
+      "loss": 2.5585,
+      "step": 1429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857458608509064e-05,
+      "loss": 2.506,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885557507201886e-05,
+      "loss": 2.5156,
+      "step": 1431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853690078481307e-05,
+      "loss": 2.4962,
+      "step": 1432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851803628206555e-05,
+      "loss": 2.6052,
+      "step": 1433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884991572150499e-05,
+      "loss": 2.527,
+      "step": 1434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848026358687236e-05,
+      "loss": 2.5497,
+      "step": 1435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884613554006416e-05,
+      "loss": 2.5448,
+      "step": 1436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884424326594686e-05,
+      "loss": 2.6331,
+      "step": 1437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842349536646694e-05,
+      "loss": 2.5741,
+      "step": 1438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840454352475235e-05,
+      "loss": 2.6316,
+      "step": 1439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883855771374431e-05,
+      "loss": 2.576,
+      "step": 1440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836659620765978e-05,
+      "loss": 2.5458,
+      "step": 1441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834760073852547e-05,
+      "loss": 2.602,
+      "step": 1442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883285907331655e-05,
+      "loss": 2.5508,
+      "step": 1443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830956619470776e-05,
+      "loss": 2.4208,
+      "step": 1444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829052712628234e-05,
+      "loss": 2.5466,
+      "step": 1445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882714735310219e-05,
+      "loss": 2.556,
+      "step": 1446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825240541206142e-05,
+      "loss": 2.5978,
+      "step": 1447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882333227725382e-05,
+      "loss": 2.546,
+      "step": 1448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821422561559208e-05,
+      "loss": 2.5306,
+      "step": 1449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819511394436506e-05,
+      "loss": 2.5665,
+      "step": 1450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881759877620018e-05,
+      "loss": 2.6672,
+      "step": 1451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815684707164917e-05,
+      "loss": 2.5422,
+      "step": 1452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881376918764565e-05,
+      "loss": 2.6732,
+      "step": 1453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811852217957543e-05,
+      "loss": 2.5969,
+      "step": 1454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880993379841601e-05,
+      "loss": 2.5813,
+      "step": 1455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808013929336685e-05,
+      "loss": 2.5662,
+      "step": 1456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806092611035468e-05,
+      "loss": 2.5364,
+      "step": 1457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880416984382847e-05,
+      "loss": 2.5094,
+      "step": 1458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880224562803206e-05,
+      "loss": 2.5662,
+      "step": 1459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800319963962836e-05,
+      "loss": 2.5593,
+      "step": 1460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8798392851937635e-05,
+      "loss": 2.6037,
+      "step": 1461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879646429227353e-05,
+      "loss": 2.5477,
+      "step": 1462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8794534285287843e-05,
+      "loss": 2.4649,
+      "step": 1463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792602831298127e-05,
+      "loss": 2.5308,
+      "step": 1464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879066993062216e-05,
+      "loss": 2.5424,
+      "step": 1465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788735583577983e-05,
+      "loss": 2.6161,
+      "step": 1466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878679979048386e-05,
+      "loss": 2.6253,
+      "step": 1467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784862551658298e-05,
+      "loss": 2.6231,
+      "step": 1468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878292386742003e-05,
+      "loss": 2.5614,
+      "step": 1469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878098373808805e-05,
+      "loss": 2.5876,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8779042163981566e-05,
+      "loss": 2.54,
+      "step": 1471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777099145420033e-05,
+      "loss": 2.5595,
+      "step": 1472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775154682723152e-05,
+      "loss": 2.5609,
+      "step": 1473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877320877621085e-05,
+      "loss": 2.4939,
+      "step": 1474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771261426203297e-05,
+      "loss": 2.5067,
+      "step": 1475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769312633020898e-05,
+      "loss": 2.6105,
+      "step": 1476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767362396984297e-05,
+      "loss": 2.4706,
+      "step": 1477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765410718414375e-05,
+      "loss": 2.5149,
+      "step": 1478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876345759763225e-05,
+      "loss": 2.5272,
+      "step": 1479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876150303495928e-05,
+      "loss": 2.5829,
+      "step": 1480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759547030717054e-05,
+      "loss": 2.5186,
+      "step": 1481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757589585227408e-05,
+      "loss": 2.6007,
+      "step": 1482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755630698812403e-05,
+      "loss": 2.6342,
+      "step": 1483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8753670371794346e-05,
+      "loss": 2.5736,
+      "step": 1484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751708604495777e-05,
+      "loss": 2.5569,
+      "step": 1485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874974539723948e-05,
+      "loss": 2.4381,
+      "step": 1486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747780750348462e-05,
+      "loss": 2.5346,
+      "step": 1487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874581466414598e-05,
+      "loss": 2.559,
+      "step": 1488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874384713895552e-05,
+      "loss": 2.5718,
+      "step": 1489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741878175100813e-05,
+      "loss": 2.4549,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739907772905817e-05,
+      "loss": 2.501,
+      "step": 1491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737935932694732e-05,
+      "loss": 2.5046,
+      "step": 1492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735962654791994e-05,
+      "loss": 2.6394,
+      "step": 1493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733987939522276e-05,
+      "loss": 2.6855,
+      "step": 1494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732011787210483e-05,
+      "loss": 2.5272,
+      "step": 1495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730034198181767e-05,
+      "loss": 2.4977,
+      "step": 1496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728055172761502e-05,
+      "loss": 2.5177,
+      "step": 1497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872607471127531e-05,
+      "loss": 2.5043,
+      "step": 1498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724092814049048e-05,
+      "loss": 2.4583,
+      "step": 1499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722109481408797e-05,
+      "loss": 2.5655,
+      "step": 1500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720124713680894e-05,
+      "loss": 2.5418,
+      "step": 1501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718138511191895e-05,
+      "loss": 2.5085,
+      "step": 1502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716150874268603e-05,
+      "loss": 2.5171,
+      "step": 1503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714161803238052e-05,
+      "loss": 2.6362,
+      "step": 1504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871217129842751e-05,
+      "loss": 2.5092,
+      "step": 1505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710179360164486e-05,
+      "loss": 2.6503,
+      "step": 1506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870818598877672e-05,
+      "loss": 2.5806,
+      "step": 1507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706191184592192e-05,
+      "loss": 2.6106,
+      "step": 1508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704194947939118e-05,
+      "loss": 2.533,
+      "step": 1509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702197279145946e-05,
+      "loss": 2.5288,
+      "step": 1510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870019817854136e-05,
+      "loss": 2.5941,
+      "step": 1511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698197646454283e-05,
+      "loss": 2.5315,
+      "step": 1512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869619568321387e-05,
+      "loss": 2.5603,
+      "step": 1513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8694192289149516e-05,
+      "loss": 2.5225,
+      "step": 1514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692187464590844e-05,
+      "loss": 2.5549,
+      "step": 1515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869018120986772e-05,
+      "loss": 2.5584,
+      "step": 1516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868817352531024e-05,
+      "loss": 2.6054,
+      "step": 1517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868616441124874e-05,
+      "loss": 2.582,
+      "step": 1518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684153868013785e-05,
+      "loss": 2.5096,
+      "step": 1519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682141895936175e-05,
+      "loss": 2.5327,
+      "step": 1520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680128495346962e-05,
+      "loss": 2.5711,
+      "step": 1521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678113666577408e-05,
+      "loss": 2.5229,
+      "step": 1522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676097409959028e-05,
+      "loss": 2.5285,
+      "step": 1523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674079725823563e-05,
+      "loss": 2.563,
+      "step": 1524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867206061450299e-05,
+      "loss": 2.6077,
+      "step": 1525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670040076329526e-05,
+      "loss": 2.5478,
+      "step": 1526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668018111635616e-05,
+      "loss": 2.5226,
+      "step": 1527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665994720753946e-05,
+      "loss": 2.5698,
+      "step": 1528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663969904017432e-05,
+      "loss": 2.5702,
+      "step": 1529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8661943661759228e-05,
+      "loss": 2.5873,
+      "step": 1530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659915994312717e-05,
+      "loss": 2.5596,
+      "step": 1531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865788690201152e-05,
+      "loss": 2.5162,
+      "step": 1532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655856385189498e-05,
+      "loss": 2.6515,
+      "step": 1533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653824444180733e-05,
+      "loss": 2.6087,
+      "step": 1534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865179107931956e-05,
+      "loss": 2.5789,
+      "step": 1535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864975629094053e-05,
+      "loss": 2.6619,
+      "step": 1536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864772007937843e-05,
+      "loss": 2.6021,
+      "step": 1537
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645682444968302e-05,
+      "loss": 2.5572,
+      "step": 1538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86436433880454e-05,
+      "loss": 2.5072,
+      "step": 1539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641602908945215e-05,
+      "loss": 2.5833,
+      "step": 1540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639561008003482e-05,
+      "loss": 2.4884,
+      "step": 1541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8637517685556164e-05,
+      "loss": 2.5841,
+      "step": 1542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8635472941939453e-05,
+      "loss": 2.527,
+      "step": 1543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633426777489782e-05,
+      "loss": 2.5135,
+      "step": 1544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631379192543818e-05,
+      "loss": 2.6883,
+      "step": 1545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629330187438457e-05,
+      "loss": 2.6173,
+      "step": 1546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627279762510832e-05,
+      "loss": 2.5988,
+      "step": 1547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625227918098306e-05,
+      "loss": 2.4866,
+      "step": 1548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623174654538486e-05,
+      "loss": 2.5907,
+      "step": 1549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862111997216919e-05,
+      "loss": 2.5839,
+      "step": 1550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86190638713285e-05,
+      "loss": 2.4729,
+      "step": 1551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617006352354706e-05,
+      "loss": 2.4831,
+      "step": 1552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614947415586343e-05,
+      "loss": 2.5829,
+      "step": 1553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612887061362177e-05,
+      "loss": 2.5248,
+      "step": 1554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861082529002121e-05,
+      "loss": 2.5328,
+      "step": 1555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608762101902668e-05,
+      "loss": 2.5726,
+      "step": 1556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8606697497346028e-05,
+      "loss": 2.545,
+      "step": 1557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604631476690972e-05,
+      "loss": 2.6905,
+      "step": 1558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602564040277445e-05,
+      "loss": 2.544,
+      "step": 1559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860049518844561e-05,
+      "loss": 2.5314,
+      "step": 1560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598424921535858e-05,
+      "loss": 2.5552,
+      "step": 1561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8596353239888826e-05,
+      "loss": 2.5682,
+      "step": 1562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594280143845372e-05,
+      "loss": 2.6248,
+      "step": 1563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859220563374659e-05,
+      "loss": 2.6052,
+      "step": 1564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859012970993382e-05,
+      "loss": 2.5851,
+      "step": 1565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858805237274861e-05,
+      "loss": 2.535,
+      "step": 1566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858597362253276e-05,
+      "loss": 2.6007,
+      "step": 1567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583893459628298e-05,
+      "loss": 2.6122,
+      "step": 1568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8581811884377477e-05,
+      "loss": 2.51,
+      "step": 1569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579728897122787e-05,
+      "loss": 2.5213,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8577644498206958e-05,
+      "loss": 2.4741,
+      "step": 1571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575558687972944e-05,
+      "loss": 2.5666,
+      "step": 1572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573471466763928e-05,
+      "loss": 2.6135,
+      "step": 1573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8571382834923334e-05,
+      "loss": 2.6694,
+      "step": 1574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569292792794812e-05,
+      "loss": 2.5634,
+      "step": 1575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856720134072225e-05,
+      "loss": 2.511,
+      "step": 1576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565108479049754e-05,
+      "loss": 2.6285,
+      "step": 1577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856301420812168e-05,
+      "loss": 2.5734,
+      "step": 1578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560918528282607e-05,
+      "loss": 2.5582,
+      "step": 1579
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558821439877345e-05,
+      "loss": 2.444,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855672294325094e-05,
+      "loss": 2.5774,
+      "step": 1581
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855462303874866e-05,
+      "loss": 2.566,
+      "step": 1582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8552521726716023e-05,
+      "loss": 2.5308,
+      "step": 1583
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550419007498755e-05,
+      "loss": 2.3588,
+      "step": 1584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548314881442836e-05,
+      "loss": 2.6249,
+      "step": 1585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854620934889446e-05,
+      "loss": 2.5795,
+      "step": 1586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544102410200065e-05,
+      "loss": 2.5103,
+      "step": 1587
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854199406570631e-05,
+      "loss": 2.5707,
+      "step": 1588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539884315760094e-05,
+      "loss": 2.551,
+      "step": 1589
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537773160708544e-05,
+      "loss": 2.6398,
+      "step": 1590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535660600899017e-05,
+      "loss": 2.53,
+      "step": 1591
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8533546636679102e-05,
+      "loss": 2.5519,
+      "step": 1592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531431268396614e-05,
+      "loss": 2.5126,
+      "step": 1593
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852931449639961e-05,
+      "loss": 2.5084,
+      "step": 1594
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852719632103637e-05,
+      "loss": 2.6102,
+      "step": 1595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852507674265541e-05,
+      "loss": 2.6084,
+      "step": 1596
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852295576160547e-05,
+      "loss": 2.7006,
+      "step": 1597
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520833378235523e-05,
+      "loss": 2.5875,
+      "step": 1598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8518709592894776e-05,
+      "loss": 2.6452,
+      "step": 1599
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516584405932667e-05,
+      "loss": 2.7235,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514457817698857e-05,
+      "loss": 2.6379,
+      "step": 1601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512329828543253e-05,
+      "loss": 2.6447,
+      "step": 1602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510200438815973e-05,
+      "loss": 2.5681,
+      "step": 1603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508069648867382e-05,
+      "loss": 2.5412,
+      "step": 1604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505937459048062e-05,
+      "loss": 2.5816,
+      "step": 1605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8503803869708835e-05,
+      "loss": 2.5718,
+      "step": 1606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850166888120075e-05,
+      "loss": 2.5165,
+      "step": 1607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499532493875085e-05,
+      "loss": 2.5974,
+      "step": 1608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8497394708083352e-05,
+      "loss": 2.4541,
+      "step": 1609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495255524177285e-05,
+      "loss": 2.5284,
+      "step": 1610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493114942508864e-05,
+      "loss": 2.5687,
+      "step": 1611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490972963430277e-05,
+      "loss": 2.6388,
+      "step": 1612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488829587293957e-05,
+      "loss": 2.5576,
+      "step": 1613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486684814452563e-05,
+      "loss": 2.5883,
+      "step": 1614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848453864525899e-05,
+      "loss": 2.6073,
+      "step": 1615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8482391080066348e-05,
+      "loss": 2.6315,
+      "step": 1616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8480242119227993e-05,
+      "loss": 2.5311,
+      "step": 1617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84780917630975e-05,
+      "loss": 2.4326,
+      "step": 1618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8475940012028674e-05,
+      "loss": 2.6372,
+      "step": 1619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473786866375555e-05,
+      "loss": 2.6195,
+      "step": 1620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847163232649241e-05,
+      "loss": 2.5185,
+      "step": 1621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469476392733735e-05,
+      "loss": 2.5535,
+      "step": 1622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846731906545425e-05,
+      "loss": 2.6279,
+      "step": 1623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465160345008925e-05,
+      "loss": 2.5044,
+      "step": 1624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463000231752923e-05,
+      "loss": 2.5585,
+      "step": 1625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460838726041675e-05,
+      "loss": 2.5215,
+      "step": 1626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845867582823081e-05,
+      "loss": 2.6415,
+      "step": 1627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845651153867621e-05,
+      "loss": 2.5956,
+      "step": 1628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454345857733965e-05,
+      "loss": 2.4916,
+      "step": 1629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452178785760413e-05,
+      "loss": 2.6276,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450010323112107e-05,
+      "loss": 2.5119,
+      "step": 1631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447840470145836e-05,
+      "loss": 2.5579,
+      "step": 1632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8445669227218618e-05,
+      "loss": 2.5445,
+      "step": 1633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844349659468769e-05,
+      "loss": 2.5358,
+      "step": 1634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441322572910527e-05,
+      "loss": 2.6074,
+      "step": 1635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8439147162244837e-05,
+      "loss": 2.6426,
+      "step": 1636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436970363048544e-05,
+      "loss": 2.5773,
+      "step": 1637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843479217567981e-05,
+      "loss": 2.6192,
+      "step": 1638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432612600497018e-05,
+      "loss": 2.5363,
+      "step": 1639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430431637858785e-05,
+      "loss": 2.544,
+      "step": 1640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8428249288123957e-05,
+      "loss": 2.5352,
+      "step": 1641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426065551651604e-05,
+      "loss": 2.4571,
+      "step": 1642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423880428801024e-05,
+      "loss": 2.6081,
+      "step": 1643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421693919931748e-05,
+      "loss": 2.5425,
+      "step": 1644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841950602540353e-05,
+      "loss": 2.5665,
+      "step": 1645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841731674557635e-05,
+      "loss": 2.5679,
+      "step": 1646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8415126080810433e-05,
+      "loss": 2.4872,
+      "step": 1647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412934031466204e-05,
+      "loss": 2.52,
+      "step": 1648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841074059790434e-05,
+      "loss": 2.4991,
+      "step": 1649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408545780485734e-05,
+      "loss": 2.537,
+      "step": 1650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840634957957151e-05,
+      "loss": 2.6194,
+      "step": 1651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404151995523015e-05,
+      "loss": 2.6063,
+      "step": 1652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401953028701832e-05,
+      "loss": 2.4833,
+      "step": 1653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839975267946976e-05,
+      "loss": 2.5842,
+      "step": 1654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397550948188842e-05,
+      "loss": 2.5953,
+      "step": 1655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8395347835221335e-05,
+      "loss": 2.6098,
+      "step": 1656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839314334092972e-05,
+      "loss": 2.638,
+      "step": 1657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8390937465676724e-05,
+      "loss": 2.5846,
+      "step": 1658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388730209825278e-05,
+      "loss": 2.4881,
+      "step": 1659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838652157373856e-05,
+      "loss": 2.52,
+      "step": 1660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384311557779967e-05,
+      "loss": 2.482,
+      "step": 1661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382100162313117e-05,
+      "loss": 2.4395,
+      "step": 1662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379887387701862e-05,
+      "loss": 2.5635,
+      "step": 1663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8377673234310284e-05,
+      "loss": 2.6112,
+      "step": 1664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837545770250268e-05,
+      "loss": 2.4933,
+      "step": 1665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837324079264359e-05,
+      "loss": 2.4677,
+      "step": 1666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837102250509777e-05,
+      "loss": 2.5633,
+      "step": 1667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83688028402302e-05,
+      "loss": 2.5433,
+      "step": 1668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8366581798406096e-05,
+      "loss": 2.5636,
+      "step": 1669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364359379990894e-05,
+      "loss": 2.5627,
+      "step": 1670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362135585350257e-05,
+      "loss": 2.5793,
+      "step": 1671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835991041485008e-05,
+      "loss": 2.5866,
+      "step": 1672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8357683868856476e-05,
+      "loss": 2.5279,
+      "step": 1673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835545594773579e-05,
+      "loss": 2.5351,
+      "step": 1674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353226651854598e-05,
+      "loss": 2.5614,
+      "step": 1675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350995981579683e-05,
+      "loss": 2.4818,
+      "step": 1676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834876393727808e-05,
+      "loss": 2.5967,
+      "step": 1677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346530519317028e-05,
+      "loss": 2.463,
+      "step": 1678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8344295728064005e-05,
+      "loss": 2.5595,
+      "step": 1679
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8342059563886712e-05,
+      "loss": 2.5178,
+      "step": 1680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8339822027153075e-05,
+      "loss": 2.5374,
+      "step": 1681
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337583118231243e-05,
+      "loss": 2.5225,
+      "step": 1682
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83353428374896e-05,
+      "loss": 2.4847,
+      "step": 1683
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8333101185296743e-05,
+      "loss": 2.5462,
+      "step": 1684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83308581620215e-05,
+      "loss": 2.6119,
+      "step": 1685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328613768032934e-05,
+      "loss": 2.4749,
+      "step": 1686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8326368003700318e-05,
+      "loss": 2.5382,
+      "step": 1687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324120869393158e-05,
+      "loss": 2.6336,
+      "step": 1688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832187236548119e-05,
+      "loss": 2.5154,
+      "step": 1689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8319622492334363e-05,
+      "loss": 2.5832,
+      "step": 1690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317371250322867e-05,
+      "loss": 2.5294,
+      "step": 1691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315118639817104e-05,
+      "loss": 2.5177,
+      "step": 1692
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8312864661187705e-05,
+      "loss": 2.5778,
+      "step": 1693
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831060931480553e-05,
+      "loss": 2.5473,
+      "step": 1694
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308352601041658e-05,
+      "loss": 2.5568,
+      "step": 1695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83060945202674e-05,
+      "loss": 2.5122,
+      "step": 1696
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830383507285429e-05,
+      "loss": 2.517,
+      "step": 1697
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830157425917408e-05,
+      "loss": 2.5454,
+      "step": 1698
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829931207959875e-05,
+      "loss": 2.5658,
+      "step": 1699
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297048534500513e-05,
+      "loss": 2.5458,
+      "step": 1700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294783624251796e-05,
+      "loss": 2.5205,
+      "step": 1701
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8292517349225255e-05,
+      "loss": 2.6293,
+      "step": 1702
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829024970979377e-05,
+      "loss": 2.5446,
+      "step": 1703
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828798070633045e-05,
+      "loss": 2.5479,
+      "step": 1704
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828571033920862e-05,
+      "loss": 2.4689,
+      "step": 1705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283438608801837e-05,
+      "loss": 2.5361,
+      "step": 1706
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281165515483877e-05,
+      "loss": 2.6129,
+      "step": 1707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827889105962874e-05,
+      "loss": 2.5897,
+      "step": 1708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8276615241610654e-05,
+      "loss": 2.549,
+      "step": 1709
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827433806180407e-05,
+      "loss": 2.565,
+      "step": 1710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272059520583666e-05,
+      "loss": 2.5089,
+      "step": 1711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269779618324333e-05,
+      "loss": 2.4755,
+      "step": 1712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82674983554012e-05,
+      "loss": 2.5406,
+      "step": 1713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265215732189608e-05,
+      "loss": 2.6121,
+      "step": 1714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262931749065136e-05,
+      "loss": 2.497,
+      "step": 1715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8260646406403567e-05,
+      "loss": 2.6627,
+      "step": 1716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8258359704580927e-05,
+      "loss": 2.5138,
+      "step": 1717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256071643973454e-05,
+      "loss": 2.5722,
+      "step": 1718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825378222495761e-05,
+      "loss": 2.5756,
+      "step": 1719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825149144791009e-05,
+      "loss": 2.595,
+      "step": 1720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249199313207802e-05,
+      "loss": 2.5254,
+      "step": 1721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824690582122788e-05,
+      "loss": 2.5425,
+      "step": 1722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244610972347683e-05,
+      "loss": 2.5753,
+      "step": 1723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242314766944794e-05,
+      "loss": 2.5241,
+      "step": 1724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240017205397017e-05,
+      "loss": 2.5759,
+      "step": 1725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823771828808238e-05,
+      "loss": 2.5177,
+      "step": 1726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235418015379138e-05,
+      "loss": 2.4766,
+      "step": 1727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233116387665757e-05,
+      "loss": 2.5899,
+      "step": 1728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230813405320943e-05,
+      "loss": 2.5065,
+      "step": 1729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228509068723606e-05,
+      "loss": 2.538,
+      "step": 1730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226203378252903e-05,
+      "loss": 2.5547,
+      "step": 1731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223896334288186e-05,
+      "loss": 2.5776,
+      "step": 1732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822158793720905e-05,
+      "loss": 2.5457,
+      "step": 1733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8219278187395306e-05,
+      "loss": 2.4658,
+      "step": 1734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8216967085226983e-05,
+      "loss": 2.5048,
+      "step": 1735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821465463108434e-05,
+      "loss": 2.5566,
+      "step": 1736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212340825347855e-05,
+      "loss": 2.5292,
+      "step": 1737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8210025668398234e-05,
+      "loss": 2.5495,
+      "step": 1738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207709160616386e-05,
+      "loss": 2.5535,
+      "step": 1739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205391302383473e-05,
+      "loss": 2.526,
+      "step": 1740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8203072094080853e-05,
+      "loss": 2.6696,
+      "step": 1741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200751536090117e-05,
+      "loss": 2.6066,
+      "step": 1742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819842962879308e-05,
+      "loss": 2.6122,
+      "step": 1743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819610637257177e-05,
+      "loss": 2.4683,
+      "step": 1744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193781767808448e-05,
+      "loss": 2.5622,
+      "step": 1745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819145581488559e-05,
+      "loss": 2.5852,
+      "step": 1746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818912851418589e-05,
+      "loss": 2.5697,
+      "step": 1747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818679986609228e-05,
+      "loss": 2.6281,
+      "step": 1748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184469870987897e-05,
+      "loss": 2.536,
+      "step": 1749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182138529256107e-05,
+      "loss": 2.5399,
+      "step": 1750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817980584128049e-05,
+      "loss": 2.5921,
+      "step": 1751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817747180744486e-05,
+      "loss": 2.6533,
+      "step": 1752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817513642813325e-05,
+      "loss": 2.4883,
+      "step": 1753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172799703729898e-05,
+      "loss": 2.5983,
+      "step": 1754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170461634619285e-05,
+      "loss": 2.6037,
+      "step": 1755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81681222211861e-05,
+      "loss": 2.6976,
+      "step": 1756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165781463815258e-05,
+      "loss": 2.588,
+      "step": 1757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163439362891896e-05,
+      "loss": 2.5159,
+      "step": 1758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816109591880137e-05,
+      "loss": 2.5078,
+      "step": 1759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158751131929257e-05,
+      "loss": 2.5673,
+      "step": 1760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8156405002661355e-05,
+      "loss": 2.4959,
+      "step": 1761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154057531383682e-05,
+      "loss": 2.4712,
+      "step": 1762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151708718482477e-05,
+      "loss": 2.5142,
+      "step": 1763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149358564344207e-05,
+      "loss": 2.5871,
+      "step": 1764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8147007069355546e-05,
+      "loss": 2.5565,
+      "step": 1765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81446542339034e-05,
+      "loss": 2.4981,
+      "step": 1766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142300058374886e-05,
+      "loss": 2.5519,
+      "step": 1767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139944543157356e-05,
+      "loss": 2.5139,
+      "step": 1768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137587688638367e-05,
+      "loss": 2.4924,
+      "step": 1769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8135229495205705e-05,
+      "loss": 2.5361,
+      "step": 1770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132869963247378e-05,
+      "loss": 2.4233,
+      "step": 1771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81305090931516e-05,
+      "loss": 2.5448,
+      "step": 1772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8128146885306828e-05,
+      "loss": 2.5058,
+      "step": 1773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125783340101715e-05,
+      "loss": 2.5351,
+      "step": 1774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123418457925153e-05,
+      "loss": 2.5175,
+      "step": 1775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121052239166245e-05,
+      "loss": 2.5529,
+      "step": 1776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118684684214315e-05,
+      "loss": 2.5603,
+      "step": 1777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116315793458905e-05,
+      "loss": 2.564,
+      "step": 1778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8113945567289782e-05,
+      "loss": 2.5031,
+      "step": 1779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111574006096933e-05,
+      "loss": 2.442,
+      "step": 1780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8109201110270556e-05,
+      "loss": 2.5879,
+      "step": 1781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106826880201076e-05,
+      "loss": 2.4189,
+      "step": 1782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8104451316279136e-05,
+      "loss": 2.5634,
+      "step": 1783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81020744188956e-05,
+      "loss": 2.449,
+      "step": 1784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099696188441545e-05,
+      "loss": 2.5685,
+      "step": 1785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097316625308278e-05,
+      "loss": 2.6051,
+      "step": 1786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094935729887312e-05,
+      "loss": 2.5618,
+      "step": 1787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8092553502570392e-05,
+      "loss": 2.6342,
+      "step": 1788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 2.5793,
+      "step": 1789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808778505381674e-05,
+      "loss": 2.5351,
+      "step": 1790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085398833164578e-05,
+      "loss": 2.6151,
+      "step": 1791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808301128218561e-05,
+      "loss": 2.57,
+      "step": 1792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080622401272668e-05,
+      "loss": 2.5346,
+      "step": 1793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078232190818806e-05,
+      "loss": 2.5807,
+      "step": 1794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8075840651217297e-05,
+      "loss": 2.6501,
+      "step": 1795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8073447782861634e-05,
+      "loss": 2.5567,
+      "step": 1796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807105358614552e-05,
+      "loss": 2.663,
+      "step": 1797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068658061462883e-05,
+      "loss": 2.5427,
+      "step": 1798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066261209207875e-05,
+      "loss": 2.5864,
+      "step": 1799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806386302977486e-05,
+      "loss": 2.5605,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061463523558415e-05,
+      "loss": 2.4972,
+      "step": 1801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805906269095335e-05,
+      "loss": 2.4937,
+      "step": 1802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056660532354678e-05,
+      "loss": 2.6034,
+      "step": 1803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805425704815764e-05,
+      "loss": 2.6469,
+      "step": 1804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051852238757694e-05,
+      "loss": 2.48,
+      "step": 1805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804944610455051e-05,
+      "loss": 2.5598,
+      "step": 1806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804703864593198e-05,
+      "loss": 2.6179,
+      "step": 1807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8044629863298215e-05,
+      "loss": 2.4872,
+      "step": 1808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8042219757045546e-05,
+      "loss": 2.5835,
+      "step": 1809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8039808327570512e-05,
+      "loss": 2.5551,
+      "step": 1810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037395575269884e-05,
+      "loss": 2.5054,
+      "step": 1811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803498150054064e-05,
+      "loss": 2.5245,
+      "step": 1812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8032566103779972e-05,
+      "loss": 2.5881,
+      "step": 1813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030149385385303e-05,
+      "loss": 2.5852,
+      "step": 1814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8027731345754267e-05,
+      "loss": 2.5719,
+      "step": 1815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802531198528471e-05,
+      "loss": 2.4558,
+      "step": 1816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802289130437471e-05,
+      "loss": 2.5612,
+      "step": 1817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8020469303422536e-05,
+      "loss": 2.4817,
+      "step": 1818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8018045982826704e-05,
+      "loss": 2.609,
+      "step": 1819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8015621342985932e-05,
+      "loss": 2.5113,
+      "step": 1820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013195384299155e-05,
+      "loss": 2.5394,
+      "step": 1821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801076810716552e-05,
+      "loss": 2.5686,
+      "step": 1822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008339511984408e-05,
+      "loss": 2.6362,
+      "step": 1823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005909599155407e-05,
+      "loss": 2.5684,
+      "step": 1824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800347836907831e-05,
+      "loss": 2.5565,
+      "step": 1825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001045822153153e-05,
+      "loss": 2.4714,
+      "step": 1826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998611958780162e-05,
+      "loss": 2.5081,
+      "step": 1827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996176779359795e-05,
+      "loss": 2.611,
+      "step": 1828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993740284292727e-05,
+      "loss": 2.5944,
+      "step": 1829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7991302473979838e-05,
+      "loss": 2.5464,
+      "step": 1830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7988863348822237e-05,
+      "loss": 2.6297,
+      "step": 1831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7986422909221247e-05,
+      "loss": 2.4201,
+      "step": 1832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7983981155578398e-05,
+      "loss": 2.6288,
+      "step": 1833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981538088295444e-05,
+      "loss": 2.6453,
+      "step": 1834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7979093707774354e-05,
+      "loss": 2.5777,
+      "step": 1835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797664801441732e-05,
+      "loss": 2.5039,
+      "step": 1836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797420100862673e-05,
+      "loss": 2.499,
+      "step": 1837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797175269080521e-05,
+      "loss": 2.6051,
+      "step": 1838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969303061355586e-05,
+      "loss": 2.5871,
+      "step": 1839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796685212068092e-05,
+      "loss": 2.4317,
+      "step": 1840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7964399869184456e-05,
+      "loss": 2.5588,
+      "step": 1841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796194630726969e-05,
+      "loss": 2.5931,
+      "step": 1842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7959491435340313e-05,
+      "loss": 2.5225,
+      "step": 1843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7957035253800233e-05,
+      "loss": 2.5498,
+      "step": 1844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795457776305358e-05,
+      "loss": 2.6018,
+      "step": 1845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7952118963504697e-05,
+      "loss": 2.5961,
+      "step": 1846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949658855558138e-05,
+      "loss": 2.5862,
+      "step": 1847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794719743961867e-05,
+      "loss": 2.6482,
+      "step": 1848
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944734716091294e-05,
+      "loss": 2.5355,
+      "step": 1849
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7942270685381206e-05,
+      "loss": 2.6177,
+      "step": 1850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7939805347893824e-05,
+      "loss": 2.5135,
+      "step": 1851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937338704034777e-05,
+      "loss": 2.5286,
+      "step": 1852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934870754209925e-05,
+      "loss": 2.5098,
+      "step": 1853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932401498825316e-05,
+      "loss": 2.5159,
+      "step": 1854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929930938287237e-05,
+      "loss": 2.479,
+      "step": 1855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7927459073002177e-05,
+      "loss": 2.5593,
+      "step": 1856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7924985903376848e-05,
+      "loss": 2.5799,
+      "step": 1857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7922511429818165e-05,
+      "loss": 2.5407,
+      "step": 1858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920035652733263e-05,
+      "loss": 2.6529,
+      "step": 1859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917558572529503e-05,
+      "loss": 2.5419,
+      "step": 1860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791508018961444e-05,
+      "loss": 2.4603,
+      "step": 1861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912600504395857e-05,
+      "loss": 2.4558,
+      "step": 1862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791011951728175e-05,
+      "loss": 2.5269,
+      "step": 1863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790763722868032e-05,
+      "loss": 2.613,
+      "step": 1864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905153638999996e-05,
+      "loss": 2.4378,
+      "step": 1865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7902668748649413e-05,
+      "loss": 2.4638,
+      "step": 1866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790018255803742e-05,
+      "loss": 2.6113,
+      "step": 1867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789769506757308e-05,
+      "loss": 2.5665,
+      "step": 1868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895206277665675e-05,
+      "loss": 2.5974,
+      "step": 1869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789271618872469e-05,
+      "loss": 2.5619,
+      "step": 1870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890224801159834e-05,
+      "loss": 2.5481,
+      "step": 1871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7887732115381028e-05,
+      "loss": 2.589,
+      "step": 1872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.78852381317984e-05,
+      "loss": 2.5551,
+      "step": 1873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7882742850822304e-05,
+      "loss": 2.6717,
+      "step": 1874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788024627286329e-05,
+      "loss": 2.5068,
+      "step": 1875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787774839833214e-05,
+      "loss": 2.5096,
+      "step": 1876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7875249227639832e-05,
+      "loss": 2.5064,
+      "step": 1877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7872748761197573e-05,
+      "loss": 2.5569,
+      "step": 1878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7870246999416773e-05,
+      "loss": 2.6513,
+      "step": 1879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786774394270906e-05,
+      "loss": 2.5681,
+      "step": 1880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865239591486268e-05,
+      "loss": 2.6014,
+      "step": 1881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7862733946160452e-05,
+      "loss": 2.5659,
+      "step": 1882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786022700714388e-05,
+      "loss": 2.5245,
+      "step": 1883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857718774849026e-05,
+      "loss": 2.508,
+      "step": 1884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785520924968858e-05,
+      "loss": 2.6123,
+      "step": 1885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852698432075454e-05,
+      "loss": 2.5854,
+      "step": 1886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785018632242275e-05,
+      "loss": 2.5551,
+      "step": 1887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784767292114381e-05,
+      "loss": 2.5691,
+      "step": 1888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7845158228652165e-05,
+      "loss": 2.6414,
+      "step": 1889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7842642245361578e-05,
+      "loss": 2.5542,
+      "step": 1890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840124971686003e-05,
+      "loss": 2.5468,
+      "step": 1891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783760640803963e-05,
+      "loss": 2.5568,
+      "step": 1892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835086554836843e-05,
+      "loss": 2.4727,
+      "step": 1893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832565412492245e-05,
+      "loss": 2.5487,
+      "step": 1894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7830042981420652e-05,
+      "loss": 2.5521,
+      "step": 1895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827519262037093e-05,
+      "loss": 2.65,
+      "step": 1896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824994254756804e-05,
+      "loss": 2.4779,
+      "step": 1897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822467959995237e-05,
+      "loss": 2.5473,
+      "step": 1898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781994037816805e-05,
+      "loss": 2.5317,
+      "step": 1899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817411509691126e-05,
+      "loss": 2.5474,
+      "step": 1900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7814881354980547e-05,
+      "loss": 2.4489,
+      "step": 1901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7812349914452605e-05,
+      "loss": 2.5244,
+      "step": 1902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809817188523818e-05,
+      "loss": 2.5865,
+      "step": 1903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807283177610905e-05,
+      "loss": 2.5419,
+      "step": 1904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7804747882130792e-05,
+      "loss": 2.5121,
+      "step": 1905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780221130250063e-05,
+      "loss": 2.3918,
+      "step": 1906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7799673439137768e-05,
+      "loss": 2.5894,
+      "step": 1907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797134292459778e-05,
+      "loss": 2.555,
+      "step": 1908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7794593862884432e-05,
+      "loss": 2.5501,
+      "step": 1909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792052150829718e-05,
+      "loss": 2.529,
+      "step": 1910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778950915671384e-05,
+      "loss": 2.4743,
+      "step": 1911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786964880955203e-05,
+      "loss": 2.632,
+      "step": 1912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778441932397243e-05,
+      "loss": 2.5521,
+      "step": 1913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781872486184352e-05,
+      "loss": 2.6092,
+      "step": 1914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779324368010015e-05,
+      "loss": 2.5671,
+      "step": 1915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777677496986867e-05,
+      "loss": 2.5062,
+      "step": 1916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774224292179782e-05,
+      "loss": 2.5609,
+      "step": 1917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7771672335363025e-05,
+      "loss": 2.4785,
+      "step": 1918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776911909983828e-05,
+      "loss": 2.5914,
+      "step": 1919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776656458602565e-05,
+      "loss": 2.5192,
+      "step": 1920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776400879434543e-05,
+      "loss": 2.5949,
+      "step": 1921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7761451725218146e-05,
+      "loss": 2.5509,
+      "step": 1922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775889337906452e-05,
+      "loss": 2.5281,
+      "step": 1923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756333756305485e-05,
+      "loss": 2.624,
+      "step": 1924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7753772857362195e-05,
+      "loss": 2.4841,
+      "step": 1925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751210682656e-05,
+      "loss": 2.6035,
+      "step": 1926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748647232608464e-05,
+      "loss": 2.5641,
+      "step": 1927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774608250764137e-05,
+      "loss": 2.5128,
+      "step": 1928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77435165081767e-05,
+      "loss": 2.5064,
+      "step": 1929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740949234636646e-05,
+      "loss": 2.5319,
+      "step": 1930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773838068744362e-05,
+      "loss": 2.5501,
+      "step": 1931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735810867020226e-05,
+      "loss": 2.626,
+      "step": 1932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7733239773789305e-05,
+      "loss": 2.5103,
+      "step": 1933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773066740817387e-05,
+      "loss": 2.516,
+      "step": 1934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7728093770597183e-05,
+      "loss": 2.4977,
+      "step": 1935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7725518861482683e-05,
+      "loss": 2.5405,
+      "step": 1936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722942681254035e-05,
+      "loss": 2.5269,
+      "step": 1937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7720365230335114e-05,
+      "loss": 2.5621,
+      "step": 1938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7717786509149994e-05,
+      "loss": 2.5531,
+      "step": 1939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771520651812296e-05,
+      "loss": 2.5003,
+      "step": 1940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771262525767852e-05,
+      "loss": 2.502,
+      "step": 1941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7710042728241374e-05,
+      "loss": 2.5334,
+      "step": 1942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707458930236436e-05,
+      "loss": 2.4769,
+      "step": 1943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7704873864088833e-05,
+      "loss": 2.6494,
+      "step": 1944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702287530223898e-05,
+      "loss": 2.5058,
+      "step": 1945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7699699929067167e-05,
+      "loss": 2.432,
+      "step": 1946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7697111061044398e-05,
+      "loss": 2.5281,
+      "step": 1947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769452092658154e-05,
+      "loss": 2.5439,
+      "step": 1948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691929526104762e-05,
+      "loss": 2.7069,
+      "step": 1949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7689336860040445e-05,
+      "loss": 2.5275,
+      "step": 1950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768674292881516e-05,
+      "loss": 2.4783,
+      "step": 1951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7684147732855712e-05,
+      "loss": 2.5124,
+      "step": 1952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768155127258909e-05,
+      "loss": 2.4812,
+      "step": 1953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7678953548442503e-05,
+      "loss": 2.4946,
+      "step": 1954
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767635456084337e-05,
+      "loss": 2.5882,
+      "step": 1955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7673754310219313e-05,
+      "loss": 2.4911,
+      "step": 1956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767115279699816e-05,
+      "loss": 2.5473,
+      "step": 1957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668550021607948e-05,
+      "loss": 2.5392,
+      "step": 1958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766594598447693e-05,
+      "loss": 2.4456,
+      "step": 1959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663340686033556e-05,
+      "loss": 2.5653,
+      "step": 1960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7660734126706488e-05,
+      "loss": 2.5049,
+      "step": 1961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7658126306924595e-05,
+      "loss": 2.538,
+      "step": 1962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765551722711695e-05,
+      "loss": 2.6045,
+      "step": 1963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7652906887712844e-05,
+      "loss": 2.5323,
+      "step": 1964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765029528914176e-05,
+      "loss": 2.59,
+      "step": 1965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76476824318334e-05,
+      "loss": 2.4665,
+      "step": 1966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645068316217668e-05,
+      "loss": 2.4997,
+      "step": 1967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642452942724676e-05,
+      "loss": 2.5732,
+      "step": 1968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639836311784742e-05,
+      "loss": 2.6382,
+      "step": 1969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7637218423828396e-05,
+      "loss": 2.5022,
+      "step": 1970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7634599279286366e-05,
+      "loss": 2.5236,
+      "step": 1971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7631978878589597e-05,
+      "loss": 2.4889,
+      "step": 1972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762935722216923e-05,
+      "loss": 2.5048,
+      "step": 1973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7626734310456618e-05,
+      "loss": 2.5265,
+      "step": 1974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7624110143883325e-05,
+      "loss": 2.6211,
+      "step": 1975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7621484722881113e-05,
+      "loss": 2.5491,
+      "step": 1976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7618858047881956e-05,
+      "loss": 2.5598,
+      "step": 1977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7616230119318033e-05,
+      "loss": 2.6036,
+      "step": 1978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613600937621725e-05,
+      "loss": 2.4668,
+      "step": 1979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7610970503225627e-05,
+      "loss": 2.5485,
+      "step": 1980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760833881656254e-05,
+      "loss": 2.5342,
+      "step": 1981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7605705878065453e-05,
+      "loss": 2.5521,
+      "step": 1982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760307168816759e-05,
+      "loss": 2.5396,
+      "step": 1983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760043624730236e-05,
+      "loss": 2.5358,
+      "step": 1984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759779955590338e-05,
+      "loss": 2.543,
+      "step": 1985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595161614404487e-05,
+      "loss": 2.5901,
+      "step": 1986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7592522423239705e-05,
+      "loss": 2.5206,
+      "step": 1987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758988198284327e-05,
+      "loss": 2.4686,
+      "step": 1988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587240293649636e-05,
+      "loss": 2.6268,
+      "step": 1989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758459735609344e-05,
+      "loss": 2.5504,
+      "step": 1990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7581953170609542e-05,
+      "loss": 2.5172,
+      "step": 1991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579307737633003e-05,
+      "loss": 2.5146,
+      "step": 1992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7576661057599088e-05,
+      "loss": 2.5502,
+      "step": 1993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574013130943264e-05,
+      "loss": 2.5734,
+      "step": 1994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571363958101204e-05,
+      "loss": 2.5728,
+      "step": 1995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75687135395088e-05,
+      "loss": 2.5797,
+      "step": 1996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756606187560212e-05,
+      "loss": 2.5709,
+      "step": 1997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756340896681747e-05,
+      "loss": 2.4984,
+      "step": 1998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756075481359133e-05,
+      "loss": 2.551,
+      "step": 1999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558099416360415e-05,
+      "loss": 2.5804,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755544277556162e-05,
+      "loss": 2.5088,
+      "step": 2001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7552784891632056e-05,
+      "loss": 2.5034,
+      "step": 2002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755012576500904e-05,
+      "loss": 2.5712,
+      "step": 2003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547465396130084e-05,
+      "loss": 2.6956,
+      "step": 2004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754480378543291e-05,
+      "loss": 2.5192,
+      "step": 2005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542140933355453e-05,
+      "loss": 2.618,
+      "step": 2006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539476840335835e-05,
+      "loss": 2.5803,
+      "step": 2007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536811506812397e-05,
+      "loss": 2.5485,
+      "step": 2008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534144933223675e-05,
+      "loss": 2.5079,
+      "step": 2009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531477120008414e-05,
+      "loss": 2.515,
+      "step": 2010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752880806760556e-05,
+      "loss": 2.4862,
+      "step": 2011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526137776454265e-05,
+      "loss": 2.6119,
+      "step": 2012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7523466246993882e-05,
+      "loss": 2.5209,
+      "step": 2013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520793479663975e-05,
+      "loss": 2.5169,
+      "step": 2014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75181194749043e-05,
+      "loss": 2.6141,
+      "step": 2015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7515444233154826e-05,
+      "loss": 2.6155,
+      "step": 2016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751276775485572e-05,
+      "loss": 2.6691,
+      "step": 2017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751009004044736e-05,
+      "loss": 2.5674,
+      "step": 2018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7507411090370322e-05,
+      "loss": 2.5227,
+      "step": 2019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7504730905065377e-05,
+      "loss": 2.6036,
+      "step": 2020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7502049484973515e-05,
+      "loss": 2.5614,
+      "step": 2021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499366830535924e-05,
+      "loss": 2.5094,
+      "step": 2022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749668294219399e-05,
+      "loss": 2.543,
+      "step": 2023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7493997820389303e-05,
+      "loss": 2.5557,
+      "step": 2024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749131146556366e-05,
+      "loss": 2.5541,
+      "step": 2025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748862387815906e-05,
+      "loss": 2.5161,
+      "step": 2026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485935058617706e-05,
+      "loss": 2.5926,
+      "step": 2027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483245007381992e-05,
+      "loss": 2.4141,
+      "step": 2028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7480553724894535e-05,
+      "loss": 2.5574,
+      "step": 2029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477861211598138e-05,
+      "loss": 2.5729,
+      "step": 2030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7475167467935815e-05,
+      "loss": 2.5411,
+      "step": 2031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7472472494350775e-05,
+      "loss": 2.5301,
+      "step": 2032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7469776291286442e-05,
+      "loss": 2.6755,
+      "step": 2033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467078859186427e-05,
+      "loss": 2.5807,
+      "step": 2034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7464380198494555e-05,
+      "loss": 2.5262,
+      "step": 2035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7461680309654843e-05,
+      "loss": 2.6492,
+      "step": 2036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7458979193111524e-05,
+      "loss": 2.5343,
+      "step": 2037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7456276849309018e-05,
+      "loss": 2.5067,
+      "step": 2038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7453573278691957e-05,
+      "loss": 2.5756,
+      "step": 2039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450868481705172e-05,
+      "loss": 2.5335,
+      "step": 2040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744816245879369e-05,
+      "loss": 2.6003,
+      "step": 2041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7445455210402755e-05,
+      "loss": 2.4533,
+      "step": 2042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442746736977795e-05,
+      "loss": 2.5808,
+      "step": 2043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440037038964444e-05,
+      "loss": 2.6153,
+      "step": 2044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437326116808555e-05,
+      "loss": 2.4957,
+      "step": 2045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7434613970956154e-05,
+      "loss": 2.5598,
+      "step": 2046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7431900601853486e-05,
+      "loss": 2.5762,
+      "step": 2047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7429186009946997e-05,
+      "loss": 2.5724,
+      "step": 2048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7426470195683328e-05,
+      "loss": 2.4956,
+      "step": 2049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7423753159509327e-05,
+      "loss": 2.4582,
+      "step": 2050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421034901872037e-05,
+      "loss": 2.5163,
+      "step": 2051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7418315423218704e-05,
+      "loss": 2.6055,
+      "step": 2052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415594723996783e-05,
+      "loss": 2.4675,
+      "step": 2053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7412872804653918e-05,
+      "loss": 2.6027,
+      "step": 2054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410149665637953e-05,
+      "loss": 2.5613,
+      "step": 2055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7407425307396948e-05,
+      "loss": 2.5654,
+      "step": 2056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740469973037915e-05,
+      "loss": 2.6262,
+      "step": 2057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740197293503301e-05,
+      "loss": 2.5097,
+      "step": 2058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399244921807178e-05,
+      "loss": 2.5543,
+      "step": 2059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7396515691150512e-05,
+      "loss": 2.5183,
+      "step": 2060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7393785243512057e-05,
+      "loss": 2.5003,
+      "step": 2061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7391053579341068e-05,
+      "loss": 2.6332,
+      "step": 2062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388320699087004e-05,
+      "loss": 2.6315,
+      "step": 2063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738558660319951e-05,
+      "loss": 2.535,
+      "step": 2064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7382851292128445e-05,
+      "loss": 2.5586,
+      "step": 2065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7380114766323858e-05,
+      "loss": 2.5124,
+      "step": 2066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7377377026236003e-05,
+      "loss": 2.5139,
+      "step": 2067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737463807231533e-05,
+      "loss": 2.5484,
+      "step": 2068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73718979050125e-05,
+      "loss": 2.5668,
+      "step": 2069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7369156524778357e-05,
+      "loss": 2.6419,
+      "step": 2070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736641393206395e-05,
+      "loss": 2.5531,
+      "step": 2071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736367012732054e-05,
+      "loss": 2.5171,
+      "step": 2072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7360925110999568e-05,
+      "loss": 2.6051,
+      "step": 2073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358178883552694e-05,
+      "loss": 2.4773,
+      "step": 2074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355431445431755e-05,
+      "loss": 2.424,
+      "step": 2075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352682797088806e-05,
+      "loss": 2.4771,
+      "step": 2076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7349932938976096e-05,
+      "loss": 2.6651,
+      "step": 2077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734718187154607e-05,
+      "loss": 2.5131,
+      "step": 2078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734442959525137e-05,
+      "loss": 2.4529,
+      "step": 2079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734167611054484e-05,
+      "loss": 2.5056,
+      "step": 2080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7338921417879532e-05,
+      "loss": 2.5891,
+      "step": 2081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733616551770868e-05,
+      "loss": 2.4823,
+      "step": 2082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333408410485727e-05,
+      "loss": 2.6432,
+      "step": 2083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733065009666431e-05,
+      "loss": 2.4984,
+      "step": 2084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732789057669827e-05,
+      "loss": 2.3992,
+      "step": 2085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732512985104164e-05,
+      "loss": 2.5756,
+      "step": 2086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322367920148653e-05,
+      "loss": 2.563,
+      "step": 2087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731960478447375e-05,
+      "loss": 2.5171,
+      "step": 2088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7316840444471553e-05,
+      "loss": 2.5104,
+      "step": 2089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314074900596898e-05,
+      "loss": 2.4904,
+      "step": 2090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7311308153304808e-05,
+      "loss": 2.5693,
+      "step": 2091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7308540203050513e-05,
+      "loss": 2.5956,
+      "step": 2092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305771050289427e-05,
+      "loss": 2.5698,
+      "step": 2093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730300069547718e-05,
+      "loss": 2.5398,
+      "step": 2094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730022913906959e-05,
+      "loss": 2.5625,
+      "step": 2095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297456381522672e-05,
+      "loss": 2.5183,
+      "step": 2096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294682423292633e-05,
+      "loss": 2.6323,
+      "step": 2097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72919072648359e-05,
+      "loss": 2.5958,
+      "step": 2098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728913090660907e-05,
+      "loss": 2.5835,
+      "step": 2099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7286353349068948e-05,
+      "loss": 2.6225,
+      "step": 2100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728357459267255e-05,
+      "loss": 2.5368,
+      "step": 2101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280794637877066e-05,
+      "loss": 2.5619,
+      "step": 2102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72780134851399e-05,
+      "loss": 2.5383,
+      "step": 2103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7275231134918652e-05,
+      "loss": 2.518,
+      "step": 2104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72724475876711e-05,
+      "loss": 2.5261,
+      "step": 2105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7269662843855247e-05,
+      "loss": 2.5429,
+      "step": 2106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266876903929275e-05,
+      "loss": 2.5669,
+      "step": 2107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264089768351565e-05,
+      "loss": 2.5223,
+      "step": 2108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7261301437580696e-05,
+      "loss": 2.5837,
+      "step": 2109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725851191207545e-05,
+      "loss": 2.5589,
+      "step": 2110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7255721192294798e-05,
+      "loss": 2.5356,
+      "step": 2111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725292927869791e-05,
+      "loss": 2.6185,
+      "step": 2112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725013617174415e-05,
+      "loss": 2.5729,
+      "step": 2113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724734187189308e-05,
+      "loss": 2.5161,
+      "step": 2114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7244546379604463e-05,
+      "loss": 2.5567,
+      "step": 2115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7241749695338246e-05,
+      "loss": 2.4207,
+      "step": 2116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238951819554587e-05,
+      "loss": 2.595,
+      "step": 2117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236152752713828e-05,
+      "loss": 2.5876,
+      "step": 2118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7233352495276517e-05,
+      "loss": 2.5549,
+      "step": 2119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723055104770339e-05,
+      "loss": 2.6093,
+      "step": 2120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227748410455384e-05,
+      "loss": 2.5558,
+      "step": 2121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7224944583993627e-05,
+      "loss": 2.6532,
+      "step": 2122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7222139568779445e-05,
+      "loss": 2.5504,
+      "step": 2123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7219333365274363e-05,
+      "loss": 2.5579,
+      "step": 2124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721652597394009e-05,
+      "loss": 2.5006,
+      "step": 2125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721371739523855e-05,
+      "loss": 2.4972,
+      "step": 2126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210907629631842e-05,
+      "loss": 2.5166,
+      "step": 2127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208096677582276e-05,
+      "loss": 2.6473,
+      "step": 2128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7205284539552346e-05,
+      "loss": 2.5116,
+      "step": 2129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7202471216004742e-05,
+      "loss": 2.4608,
+      "step": 2130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199656707402365e-05,
+      "loss": 2.5674,
+      "step": 2131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7196841014208286e-05,
+      "loss": 2.6021,
+      "step": 2132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719402413688579e-05,
+      "loss": 2.5764,
+      "step": 2133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719120607589835e-05,
+      "loss": 2.535,
+      "step": 2134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718838683170963e-05,
+      "loss": 2.5224,
+      "step": 2135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71855664047835e-05,
+      "loss": 2.5719,
+      "step": 2136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718274479558401e-05,
+      "loss": 2.5698,
+      "step": 2137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717992200457542e-05,
+      "loss": 2.4378,
+      "step": 2138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717709803222217e-05,
+      "loss": 2.57,
+      "step": 2139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71742728789889e-05,
+      "loss": 2.6034,
+      "step": 2140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717144654534045e-05,
+      "loss": 2.5694,
+      "step": 2141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716861903174185e-05,
+      "loss": 2.491,
+      "step": 2142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7165790338658318e-05,
+      "loss": 2.686,
+      "step": 2143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7162960466555274e-05,
+      "loss": 2.4934,
+      "step": 2144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160129415898332e-05,
+      "loss": 2.5545,
+      "step": 2145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7157297187153293e-05,
+      "loss": 2.5796,
+      "step": 2146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154463780786157e-05,
+      "loss": 2.4979,
+      "step": 2147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715162919726312e-05,
+      "loss": 2.4876,
+      "step": 2148
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7148793437050564e-05,
+      "loss": 2.5124,
+      "step": 2149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145956500615074e-05,
+      "loss": 2.5882,
+      "step": 2150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7143118388423425e-05,
+      "loss": 2.4743,
+      "step": 2151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714027910094258e-05,
+      "loss": 2.5292,
+      "step": 2152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71374386386397e-05,
+      "loss": 2.5082,
+      "step": 2153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7134597001982144e-05,
+      "loss": 2.5544,
+      "step": 2154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131754191437447e-05,
+      "loss": 2.5309,
+      "step": 2155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7128910207473367e-05,
+      "loss": 2.529,
+      "step": 2156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7126065050557824e-05,
+      "loss": 2.632,
+      "step": 2157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7123218721158943e-05,
+      "loss": 2.6055,
+      "step": 2158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7120371219745056e-05,
+      "loss": 2.4911,
+      "step": 2159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117522546784667e-05,
+      "loss": 2.5417,
+      "step": 2160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711467270274648e-05,
+      "loss": 2.5549,
+      "step": 2161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7111821688099394e-05,
+      "loss": 2.5389,
+      "step": 2162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71089695033125e-05,
+      "loss": 2.4887,
+      "step": 2163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106116148855082e-05,
+      "loss": 2.5362,
+      "step": 2164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7103261625196607e-05,
+      "loss": 2.575,
+      "step": 2165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100405932806756e-05,
+      "loss": 2.4933,
+      "step": 2166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7097549072155377e-05,
+      "loss": 2.5459,
+      "step": 2167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709469104371253e-05,
+      "loss": 2.5166,
+      "step": 2168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7091831847948454e-05,
+      "loss": 2.5547,
+      "step": 2169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088971485333588e-05,
+      "loss": 2.5685,
+      "step": 2170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086109956338558e-05,
+      "loss": 2.4947,
+      "step": 2171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083247261434186e-05,
+      "loss": 2.5996,
+      "step": 2172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7080383401091483e-05,
+      "loss": 2.5232,
+      "step": 2173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7077518375781657e-05,
+      "loss": 2.5097,
+      "step": 2174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70746521859761e-05,
+      "loss": 2.5364,
+      "step": 2175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071784832146395e-05,
+      "loss": 2.5605,
+      "step": 2176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7068916314764323e-05,
+      "loss": 2.4756,
+      "step": 2177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7066046634301857e-05,
+      "loss": 2.5102,
+      "step": 2178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063175791231156e-05,
+      "loss": 2.5894,
+      "step": 2179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7060303786024577e-05,
+      "loss": 2.506,
+      "step": 2180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705743061915466e-05,
+      "loss": 2.5917,
+      "step": 2181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7054556291094135e-05,
+      "loss": 2.5481,
+      "step": 2182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7051680802315937e-05,
+      "loss": 2.5592,
+      "step": 2183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048804153293176e-05,
+      "loss": 2.571,
+      "step": 2184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7045926344499165e-05,
+      "loss": 2.5008,
+      "step": 2185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7043047376407403e-05,
+      "loss": 2.5867,
+      "step": 2186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7040167249491572e-05,
+      "loss": 2.535,
+      "step": 2187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037285964225562e-05,
+      "loss": 2.5021,
+      "step": 2188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034403521083436e-05,
+      "loss": 2.5593,
+      "step": 2189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7031519920539462e-05,
+      "loss": 2.5168,
+      "step": 2190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7028635163068087e-05,
+      "loss": 2.5307,
+      "step": 2191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7025749249143953e-05,
+      "loss": 2.5525,
+      "step": 2192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7022862179241894e-05,
+      "loss": 2.5905,
+      "step": 2193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019973953836933e-05,
+      "loss": 2.5789,
+      "step": 2194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7017084573404283e-05,
+      "loss": 2.6079,
+      "step": 2195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7014194038419342e-05,
+      "loss": 2.5576,
+      "step": 2196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701130234935771e-05,
+      "loss": 2.5448,
+      "step": 2197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700840950669516e-05,
+      "loss": 2.5013,
+      "step": 2198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7005515510907674e-05,
+      "loss": 2.5768,
+      "step": 2199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002620362471406e-05,
+      "loss": 2.5058,
+      "step": 2200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999724061862717e-05,
+      "loss": 2.5514,
+      "step": 2201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699682660955814e-05,
+      "loss": 2.4815,
+      "step": 2202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699392800603441e-05,
+      "loss": 2.4698,
+      "step": 2203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699102825176844e-05,
+      "loss": 2.5519,
+      "step": 2204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988127347237353e-05,
+      "loss": 2.5417,
+      "step": 2205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6985225292918433e-05,
+      "loss": 2.5316,
+      "step": 2206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6982322089289184e-05,
+      "loss": 2.5293,
+      "step": 2207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979417736827268e-05,
+      "loss": 2.5466,
+      "step": 2208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697651223601056e-05,
+      "loss": 2.513,
+      "step": 2209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697360558731711e-05,
+      "loss": 2.5149,
+      "step": 2210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6970697791225167e-05,
+      "loss": 2.5692,
+      "step": 2211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696778884821316e-05,
+      "loss": 2.5616,
+      "step": 2212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6964878758759707e-05,
+      "loss": 2.6159,
+      "step": 2213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696196752334363e-05,
+      "loss": 2.5148,
+      "step": 2214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6959055142443913e-05,
+      "loss": 2.6172,
+      "step": 2215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956141616539755e-05,
+      "loss": 2.471,
+      "step": 2216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6953226946110523e-05,
+      "loss": 2.4931,
+      "step": 2217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6950311131635787e-05,
+      "loss": 2.5408,
+      "step": 2218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6947394173595296e-05,
+      "loss": 2.6297,
+      "step": 2219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694447607246899e-05,
+      "loss": 2.5433,
+      "step": 2220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941556828736995e-05,
+      "loss": 2.4892,
+      "step": 2221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693863644287963e-05,
+      "loss": 2.6774,
+      "step": 2222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69357149153774e-05,
+      "loss": 2.5173,
+      "step": 2223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6932792246710997e-05,
+      "loss": 2.5305,
+      "step": 2224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69298684373613e-05,
+      "loss": 2.5432,
+      "step": 2225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6926943487809374e-05,
+      "loss": 2.4911,
+      "step": 2226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6924017398536475e-05,
+      "loss": 2.5699,
+      "step": 2227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6921090170024052e-05,
+      "loss": 2.5659,
+      "step": 2228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6918161802753724e-05,
+      "loss": 2.5845,
+      "step": 2229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6915232297207315e-05,
+      "loss": 2.5728,
+      "step": 2230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6912301653866833e-05,
+      "loss": 2.4717,
+      "step": 2231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6909369873214457e-05,
+      "loss": 2.4955,
+      "step": 2232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6906436955732583e-05,
+      "loss": 2.6058,
+      "step": 2233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6903502901903762e-05,
+      "loss": 2.4957,
+      "step": 2234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690056771221076e-05,
+      "loss": 2.5949,
+      "step": 2235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68976313871365e-05,
+      "loss": 2.5783,
+      "step": 2236
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689469392716413e-05,
+      "loss": 2.5643,
+      "step": 2237
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6891755332776948e-05,
+      "loss": 2.6099,
+      "step": 2238
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888815604458457e-05,
+      "loss": 2.5482,
+      "step": 2239
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6885874742692348e-05,
+      "loss": 2.564,
+      "step": 2240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882932747962488e-05,
+      "loss": 2.5259,
+      "step": 2241
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687998962075294e-05,
+      "loss": 2.5205,
+      "step": 2242
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687704536154795e-05,
+      "loss": 2.5605,
+      "step": 2243
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874099970831955e-05,
+      "loss": 2.5666,
+      "step": 2244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687115344908956e-05,
+      "loss": 2.5741,
+      "step": 2245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6868205796805583e-05,
+      "loss": 2.507,
+      "step": 2246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6865257014465007e-05,
+      "loss": 2.6186,
+      "step": 2247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6862307102553012e-05,
+      "loss": 2.5475,
+      "step": 2248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6859356061554954e-05,
+      "loss": 2.5641,
+      "step": 2249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685640389195639e-05,
+      "loss": 2.5269,
+      "step": 2250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853450594243046e-05,
+      "loss": 2.5438,
+      "step": 2251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6850496168900846e-05,
+      "loss": 2.4668,
+      "step": 2252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684754061641589e-05,
+      "loss": 2.5572,
+      "step": 2253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684458393727448e-05,
+      "loss": 2.5074,
+      "step": 2254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6841626131963074e-05,
+      "loss": 2.4774,
+      "step": 2255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838667200968343e-05,
+      "loss": 2.5574,
+      "step": 2256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6835707144777136e-05,
+      "loss": 2.588,
+      "step": 2257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683274596387648e-05,
+      "loss": 2.6125,
+      "step": 2258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6829783658753586e-05,
+      "loss": 2.575,
+      "step": 2259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6826820229895866e-05,
+      "loss": 2.5452,
+      "step": 2260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823855677790903e-05,
+      "loss": 2.5512,
+      "step": 2261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682089000292646e-05,
+      "loss": 2.4878,
+      "step": 2262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817923205790503e-05,
+      "loss": 2.5146,
+      "step": 2263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814955286871168e-05,
+      "loss": 2.6052,
+      "step": 2264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6811986246656776e-05,
+      "loss": 2.5727,
+      "step": 2265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680901608563585e-05,
+      "loss": 2.5462,
+      "step": 2266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6806044804297067e-05,
+      "loss": 2.5412,
+      "step": 2267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803072403129314e-05,
+      "loss": 2.5427,
+      "step": 2268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800098882621653e-05,
+      "loss": 2.6027,
+      "step": 2269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679712424326333e-05,
+      "loss": 2.5908,
+      "step": 2270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794148485543775e-05,
+      "loss": 2.5702,
+      "step": 2271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6791171609952606e-05,
+      "loss": 2.5896,
+      "step": 2272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6788193616979616e-05,
+      "loss": 2.5262,
+      "step": 2273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6785214507114788e-05,
+      "loss": 2.5426,
+      "step": 2274
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782234280848295e-05,
+      "loss": 2.6203,
+      "step": 2275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6779252938670483e-05,
+      "loss": 2.4839,
+      "step": 2276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776270481071883e-05,
+      "loss": 2.5874,
+      "step": 2277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6773286908543215e-05,
+      "loss": 2.5631,
+      "step": 2278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770302221575382e-05,
+      "loss": 2.6248,
+      "step": 2279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6767316420659457e-05,
+      "loss": 2.482,
+      "step": 2280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676432950628672e-05,
+      "loss": 2.5911,
+      "step": 2281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6761341478948617e-05,
+      "loss": 2.5419,
+      "step": 2282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675835233913678e-05,
+      "loss": 2.5223,
+      "step": 2283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755362087343024e-05,
+      "loss": 2.5979,
+      "step": 2284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6752370724059352e-05,
+      "loss": 2.5538,
+      "step": 2285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749378249777947e-05,
+      "loss": 2.5108,
+      "step": 2286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674638466499117e-05,
+      "loss": 2.5286,
+      "step": 2287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674338997019157e-05,
+      "loss": 2.4636,
+      "step": 2288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674039416587188e-05,
+      "loss": 2.5172,
+      "step": 2289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737397252525013e-05,
+      "loss": 2.5667,
+      "step": 2290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6734399230644064e-05,
+      "loss": 2.5867,
+      "step": 2291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731400100722308e-05,
+      "loss": 2.5605,
+      "step": 2292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672839986325321e-05,
+      "loss": 2.5249,
+      "step": 2293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725398518730406e-05,
+      "loss": 2.6081,
+      "step": 2294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722396067647727e-05,
+      "loss": 2.6247,
+      "step": 2295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719392510499177e-05,
+      "loss": 2.6148,
+      "step": 2296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6716387847778945e-05,
+      "loss": 2.5234,
+      "step": 2297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67133820799814e-05,
+      "loss": 2.5702,
+      "step": 2298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67103752076011e-05,
+      "loss": 2.4724,
+      "step": 2299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6707367231132772e-05,
+      "loss": 2.5122,
+      "step": 2300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6704358151071335e-05,
+      "loss": 2.5865,
+      "step": 2301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6701347967911888e-05,
+      "loss": 2.481,
+      "step": 2302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6698336682149708e-05,
+      "loss": 2.5015,
+      "step": 2303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6695324294280254e-05,
+      "loss": 2.4832,
+      "step": 2304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669231080479917e-05,
+      "loss": 2.5818,
+      "step": 2305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6689296214202278e-05,
+      "loss": 2.5054,
+      "step": 2306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6686280522985584e-05,
+      "loss": 2.5856,
+      "step": 2307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668326373164527e-05,
+      "loss": 2.5056,
+      "step": 2308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6680245840677703e-05,
+      "loss": 2.5599,
+      "step": 2309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677226850579428e-05,
+      "loss": 2.5562,
+      "step": 2310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667420676184718e-05,
+      "loss": 2.3791,
+      "step": 2311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667118557497786e-05,
+      "loss": 2.4118,
+      "step": 2312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6668163290468563e-05,
+      "loss": 2.5545,
+      "step": 2313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666513990881655e-05,
+      "loss": 2.5213,
+      "step": 2314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6662115430519287e-05,
+      "loss": 2.5856,
+      "step": 2315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665908985607439e-05,
+      "loss": 2.545,
+      "step": 2316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665606318597968e-05,
+      "loss": 2.5412,
+      "step": 2317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653035420733144e-05,
+      "loss": 2.4286,
+      "step": 2318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6650006560832953e-05,
+      "loss": 2.5071,
+      "step": 2319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6646976606777462e-05,
+      "loss": 2.5512,
+      "step": 2320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66439455590652e-05,
+      "loss": 2.5479,
+      "step": 2321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640913418194878e-05,
+      "loss": 2.5158,
+      "step": 2322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637880184665394e-05,
+      "loss": 2.5488,
+      "step": 2323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6634845858975815e-05,
+      "loss": 2.544,
+      "step": 2324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663181044162539e-05,
+      "loss": 2.5348,
+      "step": 2325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6628773933113553e-05,
+      "loss": 2.5125,
+      "step": 2326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625736333939915e-05,
+      "loss": 2.5584,
+      "step": 2327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6622697644604264e-05,
+      "loss": 2.5007,
+      "step": 2328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6619657865606566e-05,
+      "loss": 2.512,
+      "step": 2329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661661699744698e-05,
+      "loss": 2.5374,
+      "step": 2330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661357504062582e-05,
+      "loss": 2.5297,
+      "step": 2331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66105319956436e-05,
+      "loss": 2.5292,
+      "step": 2332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607487863001007e-05,
+      "loss": 2.5288,
+      "step": 2333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66044426431989e-05,
+      "loss": 2.6006,
+      "step": 2334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660139633673833e-05,
+      "loss": 2.5591,
+      "step": 2335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6598348944120515e-05,
+      "loss": 2.5475,
+      "step": 2336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659530046584686e-05,
+      "loss": 2.5552,
+      "step": 2337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6592250902418938e-05,
+      "loss": 2.5237,
+      "step": 2338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6589200254338512e-05,
+      "loss": 2.5826,
+      "step": 2339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658614852210752e-05,
+      "loss": 2.5503,
+      "step": 2340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658309570622807e-05,
+      "loss": 2.4783,
+      "step": 2341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6580041807202466e-05,
+      "loss": 2.5663,
+      "step": 2342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657698682553317e-05,
+      "loss": 2.4853,
+      "step": 2343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6573930761722838e-05,
+      "loss": 2.5705,
+      "step": 2344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657087361627429e-05,
+      "loss": 2.4213,
+      "step": 2345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567815389690543e-05,
+      "loss": 2.605,
+      "step": 2346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656475608247478e-05,
+      "loss": 2.486,
+      "step": 2347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656169569513035e-05,
+      "loss": 2.5639,
+      "step": 2348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65586342281608e-05,
+      "loss": 2.554,
+      "step": 2349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555571682069847e-05,
+      "loss": 2.4848,
+      "step": 2350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655250805736139e-05,
+      "loss": 2.5622,
+      "step": 2351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654944335453949e-05,
+      "loss": 2.5647,
+      "step": 2352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6546377574108407e-05,
+      "loss": 2.4892,
+      "step": 2353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654331071657256e-05,
+      "loss": 2.5318,
+      "step": 2354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6540242782436554e-05,
+      "loss": 2.6058,
+      "step": 2355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6537173772205173e-05,
+      "loss": 2.4568,
+      "step": 2356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6534103686383374e-05,
+      "loss": 2.5245,
+      "step": 2357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6531032525476288e-05,
+      "loss": 2.4825,
+      "step": 2358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527960289989236e-05,
+      "loss": 2.5208,
+      "step": 2359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65248869804277e-05,
+      "loss": 2.5306,
+      "step": 2360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6521812597297346e-05,
+      "loss": 2.5838,
+      "step": 2361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651873714110401e-05,
+      "loss": 2.549,
+      "step": 2362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6515660612353723e-05,
+      "loss": 2.6179,
+      "step": 2363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651258301155267e-05,
+      "loss": 2.4689,
+      "step": 2364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6509504339207228e-05,
+      "loss": 2.5718,
+      "step": 2365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506424595823945e-05,
+      "loss": 2.5269,
+      "step": 2366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503343781909543e-05,
+      "loss": 2.5401,
+      "step": 2367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650026189797092e-05,
+      "loss": 2.5631,
+      "step": 2368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497178944515152e-05,
+      "loss": 2.5366,
+      "step": 2369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6494094922049494e-05,
+      "loss": 2.5662,
+      "step": 2370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6491009831081375e-05,
+      "loss": 2.5088,
+      "step": 2371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6487923672118395e-05,
+      "loss": 2.579,
+      "step": 2372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648483644566834e-05,
+      "loss": 2.605,
+      "step": 2373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481748152239156e-05,
+      "loss": 2.5712,
+      "step": 2374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647865879233898e-05,
+      "loss": 2.4637,
+      "step": 2375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6475568366476117e-05,
+      "loss": 2.5221,
+      "step": 2376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6472476875159045e-05,
+      "loss": 2.5205,
+      "step": 2377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6469384318896427e-05,
+      "loss": 2.5203,
+      "step": 2378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6466290698197093e-05,
+      "loss": 2.4471,
+      "step": 2379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6463196013570048e-05,
+      "loss": 2.6022,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6460100265524475e-05,
+      "loss": 2.6087,
+      "step": 2381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457003454569735e-05,
+      "loss": 2.5036,
+      "step": 2382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453905581215352e-05,
+      "loss": 2.4585,
+      "step": 2383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6450806645971044e-05,
+      "loss": 2.4861,
+      "step": 2384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644770664934668e-05,
+      "loss": 2.5074,
+      "step": 2385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644460559185233e-05,
+      "loss": 2.5017,
+      "step": 2386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6441503473998214e-05,
+      "loss": 2.5402,
+      "step": 2387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6438400296294742e-05,
+      "loss": 2.4567,
+      "step": 2388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6435296059252493e-05,
+      "loss": 2.56,
+      "step": 2389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643219076338222e-05,
+      "loss": 2.6091,
+      "step": 2390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6429084409194852e-05,
+      "loss": 2.5452,
+      "step": 2391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6425976997201493e-05,
+      "loss": 2.5614,
+      "step": 2392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422868527913417e-05,
+      "loss": 2.4863,
+      "step": 2393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6419759001842076e-05,
+      "loss": 2.5607,
+      "step": 2394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641664841949909e-05,
+      "loss": 2.5333,
+      "step": 2395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6413536781396266e-05,
+      "loss": 2.6029,
+      "step": 2396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6410424088045565e-05,
+      "loss": 2.5062,
+      "step": 2397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640731033995914e-05,
+      "loss": 2.5839,
+      "step": 2398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6404195537649306e-05,
+      "loss": 2.4844,
+      "step": 2399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640107968162856e-05,
+      "loss": 2.5279,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639796277240956e-05,
+      "loss": 2.5589,
+      "step": 2401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639484481050515e-05,
+      "loss": 2.5482,
+      "step": 2402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6391725796428346e-05,
+      "loss": 2.5041,
+      "step": 2403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6388605730692323e-05,
+      "loss": 2.5167,
+      "step": 2404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638548461381045e-05,
+      "loss": 2.6045,
+      "step": 2405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6382362446296253e-05,
+      "loss": 2.5251,
+      "step": 2406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6379239228663433e-05,
+      "loss": 2.4677,
+      "step": 2407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6376114961425874e-05,
+      "loss": 2.5343,
+      "step": 2408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637298964509762e-05,
+      "loss": 2.493,
+      "step": 2409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6369863280192898e-05,
+      "loss": 2.5279,
+      "step": 2410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63667358672261e-05,
+      "loss": 2.5503,
+      "step": 2411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6363607406711793e-05,
+      "loss": 2.5064,
+      "step": 2412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636047789916472e-05,
+      "loss": 2.5334,
+      "step": 2413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6357347345099785e-05,
+      "loss": 2.5386,
+      "step": 2414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6354215745032082e-05,
+      "loss": 2.5814,
+      "step": 2415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635108309947686e-05,
+      "loss": 2.5689,
+      "step": 2416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634794940894955e-05,
+      "loss": 2.431,
+      "step": 2417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6344814673965754e-05,
+      "loss": 2.5202,
+      "step": 2418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6341678895041236e-05,
+      "loss": 2.4968,
+      "step": 2419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6338542072691952e-05,
+      "loss": 2.542,
+      "step": 2420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6335404207434005e-05,
+      "loss": 2.5977,
+      "step": 2421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633226529978369e-05,
+      "loss": 2.5391,
+      "step": 2422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6329125350257466e-05,
+      "loss": 2.5772,
+      "step": 2423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6325984359371955e-05,
+      "loss": 2.534,
+      "step": 2424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632284232764397e-05,
+      "loss": 2.4865,
+      "step": 2425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631969925559047e-05,
+      "loss": 2.4838,
+      "step": 2426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631655514372861e-05,
+      "loss": 2.5296,
+      "step": 2427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6313409992575698e-05,
+      "loss": 2.5384,
+      "step": 2428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631026380264922e-05,
+      "loss": 2.5462,
+      "step": 2429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630711657446684e-05,
+      "loss": 2.5017,
+      "step": 2430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630396830854638e-05,
+      "loss": 2.5658,
+      "step": 2431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6300819005405836e-05,
+      "loss": 2.6415,
+      "step": 2432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629766866556338e-05,
+      "loss": 2.5349,
+      "step": 2433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6294517289537353e-05,
+      "loss": 2.609,
+      "step": 2434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629136487784626e-05,
+      "loss": 2.5193,
+      "step": 2435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6288211431008786e-05,
+      "loss": 2.4539,
+      "step": 2436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6285056949543782e-05,
+      "loss": 2.5486,
+      "step": 2437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6281901433970263e-05,
+      "loss": 2.5285,
+      "step": 2438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627874488480743e-05,
+      "loss": 2.5076,
+      "step": 2439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275587302574634e-05,
+      "loss": 2.4988,
+      "step": 2440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6272428687791414e-05,
+      "loss": 2.5669,
+      "step": 2441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6269269040977464e-05,
+      "loss": 2.5592,
+      "step": 2442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626610836265266e-05,
+      "loss": 2.5613,
+      "step": 2443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6262946653337036e-05,
+      "loss": 2.5418,
+      "step": 2444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6259783913550814e-05,
+      "loss": 2.5014,
+      "step": 2445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256620143814364e-05,
+      "loss": 2.5394,
+      "step": 2446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6253455344648236e-05,
+      "loss": 2.5694,
+      "step": 2447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250289516573152e-05,
+      "loss": 2.5549,
+      "step": 2448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6247122660109997e-05,
+      "loss": 2.6123,
+      "step": 2449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624395477577983e-05,
+      "loss": 2.5094,
+      "step": 2450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6240785864103878e-05,
+      "loss": 2.5403,
+      "step": 2451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6237615925603533e-05,
+      "loss": 2.4018,
+      "step": 2452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623444496080036e-05,
+      "loss": 2.5405,
+      "step": 2453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6231272970216093e-05,
+      "loss": 2.4672,
+      "step": 2454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6228099954372637e-05,
+      "loss": 2.4409,
+      "step": 2455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6224925913792055e-05,
+      "loss": 2.5124,
+      "step": 2456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622175084899659e-05,
+      "loss": 2.4741,
+      "step": 2457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218574760508653e-05,
+      "loss": 2.5754,
+      "step": 2458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215397648850815e-05,
+      "loss": 2.5855,
+      "step": 2459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6212219514545823e-05,
+      "loss": 2.4618,
+      "step": 2460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620904035811659e-05,
+      "loss": 2.527,
+      "step": 2461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620586018008619e-05,
+      "loss": 2.5328,
+      "step": 2462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6202678980977885e-05,
+      "loss": 2.4937,
+      "step": 2463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619949676131508e-05,
+      "loss": 2.5727,
+      "step": 2464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6196313521621366e-05,
+      "loss": 2.5051,
+      "step": 2465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6193129262420492e-05,
+      "loss": 2.5441,
+      "step": 2466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6189943984236384e-05,
+      "loss": 2.482,
+      "step": 2467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618675768759312e-05,
+      "loss": 2.5336,
+      "step": 2468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6183570373014967e-05,
+      "loss": 2.6233,
+      "step": 2469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6180382041026345e-05,
+      "loss": 2.4807,
+      "step": 2470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6177192692151837e-05,
+      "loss": 2.5098,
+      "step": 2471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174002326916208e-05,
+      "loss": 2.5118,
+      "step": 2472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6170810945844384e-05,
+      "loss": 2.6019,
+      "step": 2473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616761854946145e-05,
+      "loss": 2.5198,
+      "step": 2474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6164425138292673e-05,
+      "loss": 2.6005,
+      "step": 2475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616123071286348e-05,
+      "loss": 2.4916,
+      "step": 2476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6158035273699455e-05,
+      "loss": 2.5423,
+      "step": 2477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6154838821326365e-05,
+      "loss": 2.5046,
+      "step": 2478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151641356270133e-05,
+      "loss": 2.5331,
+      "step": 2479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148442879056855e-05,
+      "loss": 2.4912,
+      "step": 2480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6145243390212792e-05,
+      "loss": 2.6263,
+      "step": 2481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6142042890264364e-05,
+      "loss": 2.4807,
+      "step": 2482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613884137973817e-05,
+      "loss": 2.5222,
+      "step": 2483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6135638859160967e-05,
+      "loss": 2.5512,
+      "step": 2484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6132435329059676e-05,
+      "loss": 2.4948,
+      "step": 2485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129230789961395e-05,
+      "loss": 2.4944,
+      "step": 2486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6126025242393375e-05,
+      "loss": 2.5256,
+      "step": 2487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6122818686883042e-05,
+      "loss": 2.5849,
+      "step": 2488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6119611123957987e-05,
+      "loss": 2.5497,
+      "step": 2489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116402554145962e-05,
+      "loss": 2.6389,
+      "step": 2490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6113192977974885e-05,
+      "loss": 2.5104,
+      "step": 2491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6109982395972847e-05,
+      "loss": 2.5352,
+      "step": 2492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6106770808668092e-05,
+      "loss": 2.5955,
+      "step": 2493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6103558216589047e-05,
+      "loss": 2.6407,
+      "step": 2494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6100344620264286e-05,
+      "loss": 2.4864,
+      "step": 2495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609713002022256e-05,
+      "loss": 2.5372,
+      "step": 2496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6093914416992782e-05,
+      "loss": 2.5523,
+      "step": 2497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090697811104026e-05,
+      "loss": 2.5786,
+      "step": 2498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6087480203085538e-05,
+      "loss": 2.4619,
+      "step": 2499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084261593466727e-05,
+      "loss": 2.5222,
+      "step": 2500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6081041982777156e-05,
+      "loss": 2.5512,
+      "step": 2501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6077821371546575e-05,
+      "loss": 2.4754,
+      "step": 2502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6074599760304872e-05,
+      "loss": 2.5164,
+      "step": 2503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6071377149582126e-05,
+      "loss": 2.5408,
+      "step": 2504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606815353990856e-05,
+      "loss": 2.5881,
+      "step": 2505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064928931814568e-05,
+      "loss": 2.5888,
+      "step": 2506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606170332583072e-05,
+      "loss": 2.5306,
+      "step": 2507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6058476722487723e-05,
+      "loss": 2.5315,
+      "step": 2508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6055249122316473e-05,
+      "loss": 2.4825,
+      "step": 2509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605202052584802e-05,
+      "loss": 2.4834,
+      "step": 2510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6048790933613586e-05,
+      "loss": 2.5652,
+      "step": 2511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6045560346144542e-05,
+      "loss": 2.4594,
+      "step": 2512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6042328763972436e-05,
+      "loss": 2.4743,
+      "step": 2513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6039096187628968e-05,
+      "loss": 2.5256,
+      "step": 2514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6035862617646013e-05,
+      "loss": 2.5093,
+      "step": 2515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6032628054555606e-05,
+      "loss": 2.5455,
+      "step": 2516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6029392498889942e-05,
+      "loss": 2.5431,
+      "step": 2517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602615595118138e-05,
+      "loss": 2.5561,
+      "step": 2518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602291841196245e-05,
+      "loss": 2.5327,
+      "step": 2519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601967988176583e-05,
+      "loss": 2.5532,
+      "step": 2520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6016440361124373e-05,
+      "loss": 2.4874,
+      "step": 2521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013199850571097e-05,
+      "loss": 2.4473,
+      "step": 2522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6009958350639174e-05,
+      "loss": 2.5301,
+      "step": 2523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600671586186194e-05,
+      "loss": 2.5265,
+      "step": 2524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60034723847729e-05,
+      "loss": 2.6413,
+      "step": 2525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000227919905716e-05,
+      "loss": 2.5732,
+      "step": 2526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599698246779421e-05,
+      "loss": 2.6374,
+      "step": 2527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599373602897238e-05,
+      "loss": 2.5784,
+      "step": 2528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599048860397437e-05,
+      "loss": 2.52,
+      "step": 2529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5987240193334495e-05,
+      "loss": 2.6052,
+      "step": 2530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5983990797587234e-05,
+      "loss": 2.6068,
+      "step": 2531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598074041726722e-05,
+      "loss": 2.4175,
+      "step": 2532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5977489052909253e-05,
+      "loss": 2.5367,
+      "step": 2533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59742367050483e-05,
+      "loss": 2.5918,
+      "step": 2534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597098337421947e-05,
+      "loss": 2.4677,
+      "step": 2535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596772906095807e-05,
+      "loss": 2.4953,
+      "step": 2536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596447376579953e-05,
+      "loss": 2.4752,
+      "step": 2537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596121748927946e-05,
+      "loss": 2.578,
+      "step": 2538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595796023193364e-05,
+      "loss": 2.5325,
+      "step": 2539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5954701994297988e-05,
+      "loss": 2.4649,
+      "step": 2540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5951442776908607e-05,
+      "loss": 2.5178,
+      "step": 2541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5948182580301742e-05,
+      "loss": 2.5738,
+      "step": 2542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5944921405013813e-05,
+      "loss": 2.5699,
+      "step": 2543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5941659251581394e-05,
+      "loss": 2.5427,
+      "step": 2544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938396120541224e-05,
+      "loss": 2.5952,
+      "step": 2545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59351320124302e-05,
+      "loss": 2.5154,
+      "step": 2546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5931866927785375e-05,
+      "loss": 2.5931,
+      "step": 2547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5928600867143975e-05,
+      "loss": 2.527,
+      "step": 2548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5925333831043373e-05,
+      "loss": 2.4253,
+      "step": 2549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5922065820021113e-05,
+      "loss": 2.5233,
+      "step": 2550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59187968346149e-05,
+      "loss": 2.5871,
+      "step": 2551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5915526875362582e-05,
+      "loss": 2.5198,
+      "step": 2552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912255942802196e-05,
+      "loss": 2.6216,
+      "step": 2553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590898403747191e-05,
+      "loss": 2.5146,
+      "step": 2554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590571115991007e-05,
+      "loss": 2.546,
+      "step": 2555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5902437310655178e-05,
+      "loss": 2.5945,
+      "step": 2556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5899162490245896e-05,
+      "loss": 2.4588,
+      "step": 2557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895886699221042e-05,
+      "loss": 2.6019,
+      "step": 2558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58926099381196e-05,
+      "loss": 2.5354,
+      "step": 2559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5889332207480705e-05,
+      "loss": 2.5771,
+      "step": 2560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588605350784366e-05,
+      "loss": 2.4916,
+      "step": 2561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5882773839747924e-05,
+      "loss": 2.531,
+      "step": 2562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5879493203733115e-05,
+      "loss": 2.543,
+      "step": 2563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587621160033901e-05,
+      "loss": 2.5683,
+      "step": 2564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587292903010555e-05,
+      "loss": 2.5127,
+      "step": 2565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5869645493572825e-05,
+      "loss": 2.529,
+      "step": 2566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5866360991281095e-05,
+      "loss": 2.5425,
+      "step": 2567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5863075523770767e-05,
+      "loss": 2.5171,
+      "step": 2568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5859789091582422e-05,
+      "loss": 2.4493,
+      "step": 2569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5856501695256785e-05,
+      "loss": 2.5949,
+      "step": 2570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5853213335334748e-05,
+      "loss": 2.5679,
+      "step": 2571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5849924012357365e-05,
+      "loss": 2.5598,
+      "step": 2572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5846633726865835e-05,
+      "loss": 2.4681,
+      "step": 2573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584334247940153e-05,
+      "loss": 2.6658,
+      "step": 2574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584005027050596e-05,
+      "loss": 2.5625,
+      "step": 2575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836757100720826e-05,
+      "loss": 2.5271,
+      "step": 2576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5833462970587955e-05,
+      "loss": 2.5438,
+      "step": 2577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583016788064935e-05,
+      "loss": 2.5153,
+      "step": 2578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5826871831447165e-05,
+      "loss": 2.4708,
+      "step": 2579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823574823523715e-05,
+      "loss": 2.5186,
+      "step": 2580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5820276857421474e-05,
+      "loss": 2.4262,
+      "step": 2581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5816977933683064e-05,
+      "loss": 2.6369,
+      "step": 2582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581367805285128e-05,
+      "loss": 2.5659,
+      "step": 2583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581037721546905e-05,
+      "loss": 2.4849,
+      "step": 2584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58070754220795e-05,
+      "loss": 2.4569,
+      "step": 2585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5803772673225872e-05,
+      "loss": 2.6207,
+      "step": 2586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800468969451585e-05,
+      "loss": 2.4573,
+      "step": 2587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5797164311300214e-05,
+      "loss": 2.5306,
+      "step": 2588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579385869931549e-05,
+      "loss": 2.4857,
+      "step": 2589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57905521340413e-05,
+      "loss": 2.5679,
+      "step": 2590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5787244616021683e-05,
+      "loss": 2.5242,
+      "step": 2591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5783936145800843e-05,
+      "loss": 2.5806,
+      "step": 2592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578062672392314e-05,
+      "loss": 2.6339,
+      "step": 2593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577731635093308e-05,
+      "loss": 2.4352,
+      "step": 2594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5774005027375348e-05,
+      "loss": 2.585,
+      "step": 2595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5770692753794755e-05,
+      "loss": 2.5116,
+      "step": 2596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5767379530736296e-05,
+      "loss": 2.5083,
+      "step": 2597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5764065358745106e-05,
+      "loss": 2.5008,
+      "step": 2598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5760750238366476e-05,
+      "loss": 2.527,
+      "step": 2599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5757434170145862e-05,
+      "loss": 2.6211,
+      "step": 2600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5754117154628875e-05,
+      "loss": 2.5302,
+      "step": 2601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5750799192361273e-05,
+      "loss": 2.4883,
+      "step": 2602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747480283888974e-05,
+      "loss": 2.5564,
+      "step": 2603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5744160429758058e-05,
+      "loss": 2.4777,
+      "step": 2604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574083963051475e-05,
+      "loss": 2.5204,
+      "step": 2605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737517886705442e-05,
+      "loss": 2.544,
+      "step": 2606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573419519887667e-05,
+      "loss": 2.5922,
+      "step": 2607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5730871567575137e-05,
+      "loss": 2.6198,
+      "step": 2608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727546993347688e-05,
+      "loss": 2.5968,
+      "step": 2609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5724221476741336e-05,
+      "loss": 2.5262,
+      "step": 2610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720895018303237e-05,
+      "loss": 2.6229,
+      "step": 2611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571756761858071e-05,
+      "loss": 2.5579,
+      "step": 2612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571423927812123e-05,
+      "loss": 2.5593,
+      "step": 2613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5710909997472423e-05,
+      "loss": 2.5167,
+      "step": 2614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707579777182065e-05,
+      "loss": 2.577,
+      "step": 2615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5704248617798102e-05,
+      "loss": 2.5207,
+      "step": 2616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5700916519868614e-05,
+      "loss": 2.5134,
+      "step": 2617
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5697583483941853e-05,
+      "loss": 2.5551,
+      "step": 2618
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5694249510566216e-05,
+      "loss": 2.5164,
+      "step": 2619
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5690914600290258e-05,
+      "loss": 2.5882,
+      "step": 2620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568757875366268e-05,
+      "loss": 2.5445,
+      "step": 2621
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5684241971232355e-05,
+      "loss": 2.4985,
+      "step": 2622
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5680904253548293e-05,
+      "loss": 2.4459,
+      "step": 2623
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567756560115966e-05,
+      "loss": 2.5486,
+      "step": 2624
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5674226014615785e-05,
+      "loss": 2.529,
+      "step": 2625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567088549446614e-05,
+      "loss": 2.5725,
+      "step": 2626
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5667544041260363e-05,
+      "loss": 2.4457,
+      "step": 2627
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5664201655548228e-05,
+      "loss": 2.5877,
+      "step": 2628
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5660858337879684e-05,
+      "loss": 2.5245,
+      "step": 2629
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5657514088804813e-05,
+      "loss": 2.5236,
+      "step": 2630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5654168908873866e-05,
+      "loss": 2.639,
+      "step": 2631
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565082279863723e-05,
+      "loss": 2.5198,
+      "step": 2632
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5647475758645473e-05,
+      "loss": 2.5113,
+      "step": 2633
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564412778944928e-05,
+      "loss": 2.491,
+      "step": 2634
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640778891599522e-05,
+      "loss": 2.5857,
+      "step": 2635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637429065647198e-05,
+      "loss": 2.4228,
+      "step": 2636
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5634078312143476e-05,
+      "loss": 2.5683,
+      "step": 2637
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5630726631639662e-05,
+      "loss": 2.5347,
+      "step": 2638
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562737402468723e-05,
+      "loss": 2.4939,
+      "step": 2639
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5624020491837805e-05,
+      "loss": 2.54,
+      "step": 2640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5620666033643147e-05,
+      "loss": 2.4403,
+      "step": 2641
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5617310650655183e-05,
+      "loss": 2.6172,
+      "step": 2642
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5613954343425996e-05,
+      "loss": 2.5668,
+      "step": 2643
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5610597112507806e-05,
+      "loss": 2.6095,
+      "step": 2644
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5607238958452995e-05,
+      "loss": 2.4155,
+      "step": 2645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5603879881814103e-05,
+      "loss": 2.5273,
+      "step": 2646
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5600519883143798e-05,
+      "loss": 2.502,
+      "step": 2647
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5597158962994935e-05,
+      "loss": 2.6139,
+      "step": 2648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593797121920484e-05,
+      "loss": 2.6368,
+      "step": 2649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5590434360473596e-05,
+      "loss": 2.5999,
+      "step": 2650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587070679207555e-05,
+      "loss": 2.6524,
+      "step": 2651
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5583706078675802e-05,
+      "loss": 2.4922,
+      "step": 2652
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5580340559431933e-05,
+      "loss": 2.5335,
+      "step": 2653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5576974122029688e-05,
+      "loss": 2.5249,
+      "step": 2654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557360676702297e-05,
+      "loss": 2.5357,
+      "step": 2655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5570238494965816e-05,
+      "loss": 2.5003,
+      "step": 2656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5566869306412424e-05,
+      "loss": 2.4847,
+      "step": 2657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556349920191715e-05,
+      "loss": 2.5846,
+      "step": 2658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556012818203448e-05,
+      "loss": 2.4543,
+      "step": 2659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5556756247319075e-05,
+      "loss": 2.562,
+      "step": 2660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5553383398325723e-05,
+      "loss": 2.4912,
+      "step": 2661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5550009635609383e-05,
+      "loss": 2.5524,
+      "step": 2662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546634959725152e-05,
+      "loss": 2.5501,
+      "step": 2663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5543259371228278e-05,
+      "loss": 2.4107,
+      "step": 2664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5539882870674166e-05,
+      "loss": 2.4775,
+      "step": 2665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5536505458618366e-05,
+      "loss": 2.4574,
+      "step": 2666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5533127135616574e-05,
+      "loss": 2.4646,
+      "step": 2667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552974790222465e-05,
+      "loss": 2.5383,
+      "step": 2668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5526367758998584e-05,
+      "loss": 2.5434,
+      "step": 2669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552298670649453e-05,
+      "loss": 2.5948,
+      "step": 2670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551960474526879e-05,
+      "loss": 2.5587,
+      "step": 2671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551622187587781e-05,
+      "loss": 2.5949,
+      "step": 2672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5512838098878193e-05,
+      "loss": 2.5512,
+      "step": 2673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.550945341482668e-05,
+      "loss": 2.6329,
+      "step": 2674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550606782428018e-05,
+      "loss": 2.5155,
+      "step": 2675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550268132779572e-05,
+      "loss": 2.4066,
+      "step": 2676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5499293925930516e-05,
+      "loss": 2.4797,
+      "step": 2677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5495905619241897e-05,
+      "loss": 2.5518,
+      "step": 2678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549251640828737e-05,
+      "loss": 2.5086,
+      "step": 2679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5489126293624566e-05,
+      "loss": 2.4817,
+      "step": 2680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5485735275811277e-05,
+      "loss": 2.5127,
+      "step": 2681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5482343355405447e-05,
+      "loss": 2.461,
+      "step": 2682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5478950532965157e-05,
+      "loss": 2.5903,
+      "step": 2683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475556809048654e-05,
+      "loss": 2.5044,
+      "step": 2684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472162184214314e-05,
+      "loss": 2.5162,
+      "step": 2685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546876665902067e-05,
+      "loss": 2.4965,
+      "step": 2686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546537023402641e-05,
+      "loss": 2.5327,
+      "step": 2687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5461972909790356e-05,
+      "loss": 2.5291,
+      "step": 2688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5458574686871482e-05,
+      "loss": 2.5149,
+      "step": 2689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545517556582892e-05,
+      "loss": 2.575,
+      "step": 2690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5451775547221944e-05,
+      "loss": 2.5552,
+      "step": 2691
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544837463160997e-05,
+      "loss": 2.5316,
+      "step": 2692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544497281955256e-05,
+      "loss": 2.498,
+      "step": 2693
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5441570111609445e-05,
+      "loss": 2.5082,
+      "step": 2694
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543816650834047e-05,
+      "loss": 2.4755,
+      "step": 2695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543476201030566e-05,
+      "loss": 2.5775,
+      "step": 2696
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543135661806516e-05,
+      "loss": 2.5137,
+      "step": 2697
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5427950332179283e-05,
+      "loss": 2.4397,
+      "step": 2698
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542454315320848e-05,
+      "loss": 2.5465,
+      "step": 2699
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5421135081713346e-05,
+      "loss": 2.5717,
+      "step": 2700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5417726118254626e-05,
+      "loss": 2.4093,
+      "step": 2701
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5414316263393216e-05,
+      "loss": 2.5803,
+      "step": 2702
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5410905517690148e-05,
+      "loss": 2.5229,
+      "step": 2703
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5407493881706616e-05,
+      "loss": 2.5151,
+      "step": 2704
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540408135600394e-05,
+      "loss": 2.4431,
+      "step": 2705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540066794114361e-05,
+      "loss": 2.5275,
+      "step": 2706
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5397253637687244e-05,
+      "loss": 2.4602,
+      "step": 2707
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5393838446196613e-05,
+      "loss": 2.5966,
+      "step": 2708
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539042236723364e-05,
+      "loss": 2.5195,
+      "step": 2709
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5387005401360376e-05,
+      "loss": 2.4758,
+      "step": 2710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538358754913904e-05,
+      "loss": 2.5077,
+      "step": 2711
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538016881113198e-05,
+      "loss": 2.499,
+      "step": 2712
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53767491879017e-05,
+      "loss": 2.5722,
+      "step": 2713
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537332868001084e-05,
+      "loss": 2.499,
+      "step": 2714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5369907288022206e-05,
+      "loss": 2.5028,
+      "step": 2715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5366485012498718e-05,
+      "loss": 2.5465,
+      "step": 2716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5363061854003468e-05,
+      "loss": 2.4831,
+      "step": 2717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535963781309968e-05,
+      "loss": 2.5249,
+      "step": 2718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5356212890350734e-05,
+      "loss": 2.5357,
+      "step": 2719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5352787086320136e-05,
+      "loss": 2.6309,
+      "step": 2720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534936040157155e-05,
+      "loss": 2.5962,
+      "step": 2721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5345932836668797e-05,
+      "loss": 2.5686,
+      "step": 2722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5342504392175816e-05,
+      "loss": 2.5409,
+      "step": 2723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5339075068656705e-05,
+      "loss": 2.528,
+      "step": 2724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5335644866675716e-05,
+      "loss": 2.5148,
+      "step": 2725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533221378679723e-05,
+      "loss": 2.5643,
+      "step": 2726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328781829585776e-05,
+      "loss": 2.5819,
+      "step": 2727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5325348995606028e-05,
+      "loss": 2.5641,
+      "step": 2728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532191528542281e-05,
+      "loss": 2.4703,
+      "step": 2729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5318480699601087e-05,
+      "loss": 2.5263,
+      "step": 2730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315045238705958e-05,
+      "loss": 2.4852,
+      "step": 2731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5311608903302684e-05,
+      "loss": 2.5846,
+      "step": 2732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5308171693956656e-05,
+      "loss": 2.453,
+      "step": 2733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5304733611233415e-05,
+      "loss": 2.6063,
+      "step": 2734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5301294655698642e-05,
+      "loss": 2.559,
+      "step": 2735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5297854827918167e-05,
+      "loss": 2.4103,
+      "step": 2736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294414128457956e-05,
+      "loss": 2.5147,
+      "step": 2737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529097255788413e-05,
+      "loss": 2.458,
+      "step": 2738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5287530116762937e-05,
+      "loss": 2.6119,
+      "step": 2739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528408680566078e-05,
+      "loss": 2.4176,
+      "step": 2740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5280642625144205e-05,
+      "loss": 2.5193,
+      "step": 2741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5277197575779897e-05,
+      "loss": 2.5676,
+      "step": 2742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5273751658134692e-05,
+      "loss": 2.5205,
+      "step": 2743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527030487277555e-05,
+      "loss": 2.4352,
+      "step": 2744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526685722026959e-05,
+      "loss": 2.5874,
+      "step": 2745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526340870118407e-05,
+      "loss": 2.5874,
+      "step": 2746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5259959316086398e-05,
+      "loss": 2.5419,
+      "step": 2747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5256509065544106e-05,
+      "loss": 2.4895,
+      "step": 2748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5253057950124884e-05,
+      "loss": 2.3806,
+      "step": 2749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524960597039656e-05,
+      "loss": 2.61,
+      "step": 2750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52461531269271e-05,
+      "loss": 2.5311,
+      "step": 2751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524269942028462e-05,
+      "loss": 2.5391,
+      "step": 2752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5239244851037374e-05,
+      "loss": 2.4769,
+      "step": 2753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5235789419753757e-05,
+      "loss": 2.4777,
+      "step": 2754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.52323331270023e-05,
+      "loss": 2.5119,
+      "step": 2755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5228875973351693e-05,
+      "loss": 2.615,
+      "step": 2756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522541795937075e-05,
+      "loss": 2.4246,
+      "step": 2757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5221959085628433e-05,
+      "loss": 2.5473,
+      "step": 2758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5218499352693852e-05,
+      "loss": 2.4827,
+      "step": 2759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5215038761136247e-05,
+      "loss": 2.5688,
+      "step": 2760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211577311525005e-05,
+      "loss": 2.5324,
+      "step": 2761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5208115004429656e-05,
+      "loss": 2.5763,
+      "step": 2762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5204651840419864e-05,
+      "loss": 2.5034,
+      "step": 2763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201187820065448e-05,
+      "loss": 2.4991,
+      "step": 2764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5197722943936348e-05,
+      "loss": 2.5875,
+      "step": 2765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194257212602666e-05,
+      "loss": 2.5454,
+      "step": 2766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5190790626634627e-05,
+      "loss": 2.5239,
+      "step": 2767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5187323186602606e-05,
+      "loss": 2.4625,
+      "step": 2768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5183854893077114e-05,
+      "loss": 2.5006,
+      "step": 2769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180385746628811e-05,
+      "loss": 2.4886,
+      "step": 2770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5176915747828487e-05,
+      "loss": 2.4286,
+      "step": 2771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173444897247076e-05,
+      "loss": 2.498,
+      "step": 2772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169973195455651e-05,
+      "loss": 2.5292,
+      "step": 2773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166500643025432e-05,
+      "loss": 2.5696,
+      "step": 2774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516302724052777e-05,
+      "loss": 2.4687,
+      "step": 2775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5159552988534158e-05,
+      "loss": 2.6135,
+      "step": 2776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5156077887616238e-05,
+      "loss": 2.4341,
+      "step": 2777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5152601938345773e-05,
+      "loss": 2.6008,
+      "step": 2778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5149125141294683e-05,
+      "loss": 2.4322,
+      "step": 2779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5145647497035016e-05,
+      "loss": 2.6007,
+      "step": 2780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5142169006138974e-05,
+      "loss": 2.4897,
+      "step": 2781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138689669178876e-05,
+      "loss": 2.5849,
+      "step": 2782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135209486727207e-05,
+      "loss": 2.4784,
+      "step": 2783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131728459356563e-05,
+      "loss": 2.5585,
+      "step": 2784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128246587639703e-05,
+      "loss": 2.4645,
+      "step": 2785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124763872149509e-05,
+      "loss": 2.5131,
+      "step": 2786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5121280313459012e-05,
+      "loss": 2.4463,
+      "step": 2787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5117795912141376e-05,
+      "loss": 2.4918,
+      "step": 2788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114310668769903e-05,
+      "loss": 2.5539,
+      "step": 2789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5110824583918036e-05,
+      "loss": 2.5482,
+      "step": 2790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510733765815936e-05,
+      "loss": 2.5611,
+      "step": 2791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510384989206759e-05,
+      "loss": 2.568,
+      "step": 2792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5100361286216587e-05,
+      "loss": 2.4315,
+      "step": 2793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5096871841180343e-05,
+      "loss": 2.4951,
+      "step": 2794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5093381557532998e-05,
+      "loss": 2.4978,
+      "step": 2795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508989043584882e-05,
+      "loss": 2.5223,
+      "step": 2796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086398476702217e-05,
+      "loss": 2.523,
+      "step": 2797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082905680667743e-05,
+      "loss": 2.5525,
+      "step": 2798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5079412048320075e-05,
+      "loss": 2.602,
+      "step": 2799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075917580234043e-05,
+      "loss": 2.499,
+      "step": 2800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5072422276984604e-05,
+      "loss": 2.5704,
+      "step": 2801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068926139146856e-05,
+      "loss": 2.5047,
+      "step": 2802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5065429167296035e-05,
+      "loss": 2.4974,
+      "step": 2803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061931362007513e-05,
+      "loss": 2.5357,
+      "step": 2804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5058432723856802e-05,
+      "loss": 2.4985,
+      "step": 2805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5054933253419543e-05,
+      "loss": 2.6036,
+      "step": 2806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5051432951271526e-05,
+      "loss": 2.5014,
+      "step": 2807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047931817988669e-05,
+      "loss": 2.4753,
+      "step": 2808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5044429854147022e-05,
+      "loss": 2.4491,
+      "step": 2809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504092706032279e-05,
+      "loss": 2.5707,
+      "step": 2810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5037423437092298e-05,
+      "loss": 2.5722,
+      "step": 2811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5033918985032012e-05,
+      "loss": 2.5856,
+      "step": 2812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5030413704718539e-05,
+      "loss": 2.5471,
+      "step": 2813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5026907596728612e-05,
+      "loss": 2.4843,
+      "step": 2814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023400661639112e-05,
+      "loss": 2.5011,
+      "step": 2815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019892900027052e-05,
+      "loss": 2.4941,
+      "step": 2816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5016384312469572e-05,
+      "loss": 2.606,
+      "step": 2817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012874899543964e-05,
+      "loss": 2.5713,
+      "step": 2818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500936466182764e-05,
+      "loss": 2.5645,
+      "step": 2819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005853599898162e-05,
+      "loss": 2.4328,
+      "step": 2820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002341714333218e-05,
+      "loss": 2.5738,
+      "step": 2821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499882900571063e-05,
+      "loss": 2.5992,
+      "step": 2822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4995315474608367e-05,
+      "loss": 2.5728,
+      "step": 2823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4991801121604521e-05,
+      "loss": 2.5536,
+      "step": 2824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4988285947277327e-05,
+      "loss": 2.4911,
+      "step": 2825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498476995220515e-05,
+      "loss": 2.597,
+      "step": 2826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981253136966495e-05,
+      "loss": 2.4719,
+      "step": 2827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977735502139994e-05,
+      "loss": 2.5584,
+      "step": 2828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4974217048304425e-05,
+      "loss": 2.5485,
+      "step": 2829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970697776038692e-05,
+      "loss": 2.6393,
+      "step": 2830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967177685921841e-05,
+      "loss": 2.5885,
+      "step": 2831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.496365677853304e-05,
+      "loss": 2.4837,
+      "step": 2832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4960135054451606e-05,
+      "loss": 2.4806,
+      "step": 2833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4956612514256983e-05,
+      "loss": 2.5379,
+      "step": 2834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4953089158528751e-05,
+      "loss": 2.5952,
+      "step": 2835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4949564987846618e-05,
+      "loss": 2.5783,
+      "step": 2836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4946040002790437e-05,
+      "loss": 2.5098,
+      "step": 2837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494251420394019e-05,
+      "loss": 2.4721,
+      "step": 2838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4938987591875988e-05,
+      "loss": 2.5374,
+      "step": 2839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935460167178085e-05,
+      "loss": 2.4878,
+      "step": 2840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4931931930426858e-05,
+      "loss": 2.5299,
+      "step": 2841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492840288220283e-05,
+      "loss": 2.4837,
+      "step": 2842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4924873023086645e-05,
+      "loss": 2.4611,
+      "step": 2843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4921342353659097e-05,
+      "loss": 2.5596,
+      "step": 2844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4917810874501091e-05,
+      "loss": 2.5154,
+      "step": 2845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491427858619368e-05,
+      "loss": 2.4997,
+      "step": 2846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910745489318051e-05,
+      "loss": 2.5151,
+      "step": 2847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907211584455519e-05,
+      "loss": 2.4646,
+      "step": 2848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490367687218753e-05,
+      "loss": 2.5247,
+      "step": 2849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4900141353095671e-05,
+      "loss": 2.4228,
+      "step": 2850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896605027761653e-05,
+      "loss": 2.4621,
+      "step": 2851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4893067896767326e-05,
+      "loss": 2.4839,
+      "step": 2852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4889529960694663e-05,
+      "loss": 2.538,
+      "step": 2853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4885991220125785e-05,
+      "loss": 2.4799,
+      "step": 2854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4882451675642937e-05,
+      "loss": 2.5534,
+      "step": 2855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878911327828489e-05,
+      "loss": 2.5933,
+      "step": 2856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4875370177264956e-05,
+      "loss": 2.4872,
+      "step": 2857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4871828224534974e-05,
+      "loss": 2.5138,
+      "step": 2858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4868285470221323e-05,
+      "loss": 2.6118,
+      "step": 2859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4864741914906902e-05,
+      "loss": 2.5349,
+      "step": 2860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861197559174754e-05,
+      "loss": 2.4983,
+      "step": 2861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4857652403608042e-05,
+      "loss": 2.5491,
+      "step": 2862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485410644879007e-05,
+      "loss": 2.489,
+      "step": 2863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4850559695304267e-05,
+      "loss": 2.5148,
+      "step": 2864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847012143734201e-05,
+      "loss": 2.5468,
+      "step": 2865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4843463794663561e-05,
+      "loss": 2.5073,
+      "step": 2866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4839914648676174e-05,
+      "loss": 2.465,
+      "step": 2867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836364706356002e-05,
+      "loss": 2.5068,
+      "step": 2868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4832813968287125e-05,
+      "loss": 2.5465,
+      "step": 2869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4829262435053769e-05,
+      "loss": 2.5784,
+      "step": 2870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4825710107240277e-05,
+      "loss": 2.5121,
+      "step": 2871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482215698543113e-05,
+      "loss": 2.6012,
+      "step": 2872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4818603070210948e-05,
+      "loss": 2.5113,
+      "step": 2873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815048362164464e-05,
+      "loss": 2.4974,
+      "step": 2874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4811492861876551e-05,
+      "loss": 2.5502,
+      "step": 2875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807936569932213e-05,
+      "loss": 2.4798,
+      "step": 2876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480437948691658e-05,
+      "loss": 2.5564,
+      "step": 2877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4800821613414923e-05,
+      "loss": 2.5081,
+      "step": 2878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479726295001262e-05,
+      "loss": 2.4738,
+      "step": 2879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793703497295209e-05,
+      "loss": 2.5341,
+      "step": 2880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790143255848333e-05,
+      "loss": 2.5254,
+      "step": 2881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4786582226257781e-05,
+      "loss": 2.5511,
+      "step": 2882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783020409109458e-05,
+      "loss": 2.5375,
+      "step": 2883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4779457804989412e-05,
+      "loss": 2.5609,
+      "step": 2884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4775894414483811e-05,
+      "loss": 2.5193,
+      "step": 2885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4772330238178958e-05,
+      "loss": 2.513,
+      "step": 2886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4768765276661277e-05,
+      "loss": 2.5046,
+      "step": 2887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4765199530517334e-05,
+      "loss": 2.581,
+      "step": 2888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476163300033381e-05,
+      "loss": 2.5012,
+      "step": 2889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4758065686697533e-05,
+      "loss": 2.4696,
+      "step": 2890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4754497590195435e-05,
+      "loss": 2.6022,
+      "step": 2891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4750928711414605e-05,
+      "loss": 2.4315,
+      "step": 2892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747359050942235e-05,
+      "loss": 2.5117,
+      "step": 2893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4743788609365667e-05,
+      "loss": 2.5423,
+      "step": 2894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740217387272355e-05,
+      "loss": 2.5126,
+      "step": 2895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736645385249889e-05,
+      "loss": 2.4817,
+      "step": 2896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733072603885985e-05,
+      "loss": 2.6109,
+      "step": 2897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4729499043768494e-05,
+      "loss": 2.4799,
+      "step": 2898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725924705485388e-05,
+      "loss": 2.4728,
+      "step": 2899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722349589624765e-05,
+      "loss": 2.4393,
+      "step": 2900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4718773696774857e-05,
+      "loss": 2.5369,
+      "step": 2901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4715197027524023e-05,
+      "loss": 2.4003,
+      "step": 2902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4711619582460753e-05,
+      "loss": 2.5982,
+      "step": 2903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708041362173646e-05,
+      "loss": 2.527,
+      "step": 2904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4704462367251457e-05,
+      "loss": 2.6617,
+      "step": 2905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4700882598283046e-05,
+      "loss": 2.5516,
+      "step": 2906
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4697302055857413e-05,
+      "loss": 2.5142,
+      "step": 2907
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4693720740563673e-05,
+      "loss": 2.4847,
+      "step": 2908
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4690138652991085e-05,
+      "loss": 2.4928,
+      "step": 2909
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4686555793729021e-05,
+      "loss": 2.6153,
+      "step": 2910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4682972163366984e-05,
+      "loss": 2.5123,
+      "step": 2911
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4679387762494609e-05,
+      "loss": 2.5382,
+      "step": 2912
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675802591701648e-05,
+      "loss": 2.4659,
+      "step": 2913
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4672216651577991e-05,
+      "loss": 2.5035,
+      "step": 2914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4668629942713643e-05,
+      "loss": 2.529,
+      "step": 2915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466504246569875e-05,
+      "loss": 2.5701,
+      "step": 2916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4661454221123566e-05,
+      "loss": 2.5642,
+      "step": 2917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657865209578484e-05,
+      "loss": 2.5529,
+      "step": 2918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4654275431654019e-05,
+      "loss": 2.4584,
+      "step": 2919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650684887940818e-05,
+      "loss": 2.5181,
+      "step": 2920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4647093579029642e-05,
+      "loss": 2.5159,
+      "step": 2921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4643501505511396e-05,
+      "loss": 2.5563,
+      "step": 2922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463990866797709e-05,
+      "loss": 2.502,
+      "step": 2923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636315067017872e-05,
+      "loss": 2.5251,
+      "step": 2924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632720703225016e-05,
+      "loss": 2.6082,
+      "step": 2925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629125577189916e-05,
+      "loss": 2.5232,
+      "step": 2926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4625529689504094e-05,
+      "loss": 2.5283,
+      "step": 2927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46219330407592e-05,
+      "loss": 2.6391,
+      "step": 2928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4618335631547002e-05,
+      "loss": 2.5506,
+      "step": 2929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4614737462459405e-05,
+      "loss": 2.42,
+      "step": 2930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611138534088424e-05,
+      "loss": 2.6181,
+      "step": 2931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4607538847026214e-05,
+      "loss": 2.6607,
+      "step": 2932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4603938401865042e-05,
+      "loss": 2.5434,
+      "step": 2933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460033719919731e-05,
+      "loss": 2.4767,
+      "step": 2934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4596735239615534e-05,
+      "loss": 2.4359,
+      "step": 2935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593132523712365e-05,
+      "loss": 2.5286,
+      "step": 2936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589529052080575e-05,
+      "loss": 2.5631,
+      "step": 2937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4585924825313053e-05,
+      "loss": 2.4916,
+      "step": 2938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4582319844002829e-05,
+      "loss": 2.5578,
+      "step": 2939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578714108743035e-05,
+      "loss": 2.5044,
+      "step": 2940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4575107620126945e-05,
+      "loss": 2.5576,
+      "step": 2941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457150037874795e-05,
+      "loss": 2.4183,
+      "step": 2942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4567892385199569e-05,
+      "loss": 2.5333,
+      "step": 2943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4564283640075435e-05,
+      "loss": 2.5689,
+      "step": 2944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456067414396931e-05,
+      "loss": 2.5031,
+      "step": 2945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4557063897475088e-05,
+      "loss": 2.4926,
+      "step": 2946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4553452901186778e-05,
+      "loss": 2.4656,
+      "step": 2947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549841155698506e-05,
+      "loss": 2.486,
+      "step": 2948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546228661604533e-05,
+      "loss": 2.5359,
+      "step": 2949
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4542615419499242e-05,
+      "loss": 2.5219,
+      "step": 2950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4539001429977133e-05,
+      "loss": 2.536,
+      "step": 2951
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4535386693632829e-05,
+      "loss": 2.5033,
+      "step": 2952
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531771211061082e-05,
+      "loss": 2.5217,
+      "step": 2953
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528154982856766e-05,
+      "loss": 2.5394,
+      "step": 2954
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4524538009614868e-05,
+      "loss": 2.538,
+      "step": 2955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4520920291930512e-05,
+      "loss": 2.4861,
+      "step": 2956
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4517301830398931e-05,
+      "loss": 2.4524,
+      "step": 2957
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513682625615491e-05,
+      "loss": 2.5692,
+      "step": 2958
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510062678175675e-05,
+      "loss": 2.608,
+      "step": 2959
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450644198867509e-05,
+      "loss": 2.4838,
+      "step": 2960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4502820557709463e-05,
+      "loss": 2.5014,
+      "step": 2961
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4499198385874642e-05,
+      "loss": 2.6423,
+      "step": 2962
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495575473766604e-05,
+      "loss": 2.5318,
+      "step": 2963
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449195182198144e-05,
+      "loss": 2.5255,
+      "step": 2964
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4488327431115366e-05,
+      "loss": 2.4592,
+      "step": 2965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484702301764718e-05,
+      "loss": 2.4508,
+      "step": 2966
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481076434525956e-05,
+      "loss": 2.4734,
+      "step": 2967
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4477449829995663e-05,
+      "loss": 2.5973,
+      "step": 2968
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447382248877053e-05,
+      "loss": 2.5293,
+      "step": 2969
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4470194411447395e-05,
+      "loss": 2.5556,
+      "step": 2970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4466565598623192e-05,
+      "loss": 2.497,
+      "step": 2971
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4462936050894986e-05,
+      "loss": 2.4593,
+      "step": 2972
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4459305768859968e-05,
+      "loss": 2.4597,
+      "step": 2973
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4455674753115438e-05,
+      "loss": 2.4979,
+      "step": 2974
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452043004258827e-05,
+      "loss": 2.449,
+      "step": 2975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448410522887681e-05,
+      "loss": 2.5304,
+      "step": 2976
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4444777309599673e-05,
+      "loss": 2.5556,
+      "step": 2977
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4441143364992584e-05,
+      "loss": 2.5939,
+      "step": 2978
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443750868966433e-05,
+      "loss": 2.5061,
+      "step": 2979
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4433873284212937e-05,
+      "loss": 2.5726,
+      "step": 2980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430237149236557e-05,
+      "loss": 2.4864,
+      "step": 2981
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4426600285333457e-05,
+      "loss": 2.5906,
+      "step": 2982
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4422962693102028e-05,
+      "loss": 2.4475,
+      "step": 2983
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4419324373140775e-05,
+      "loss": 2.5819,
+      "step": 2984
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415685326048337e-05,
+      "loss": 2.5683,
+      "step": 2985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441204555242345e-05,
+      "loss": 2.4938,
+      "step": 2986
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4408405052864993e-05,
+      "loss": 2.5063,
+      "step": 2987
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4404763827971948e-05,
+      "loss": 2.5619,
+      "step": 2988
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4401121878343423e-05,
+      "loss": 2.4957,
+      "step": 2989
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397479204578643e-05,
+      "loss": 2.4819,
+      "step": 2990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439383580727696e-05,
+      "loss": 2.4701,
+      "step": 2991
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439019168703783e-05,
+      "loss": 2.485,
+      "step": 2992
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438654684446084e-05,
+      "loss": 2.5313,
+      "step": 2993
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4382901280145693e-05,
+      "loss": 2.5578,
+      "step": 2994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4379254994692207e-05,
+      "loss": 2.52,
+      "step": 2995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375607988700327e-05,
+      "loss": 2.5061,
+      "step": 2996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4371960262770104e-05,
+      "loss": 2.4949,
+      "step": 2997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436831181750172e-05,
+      "loss": 2.4283,
+      "step": 2998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364662653495468e-05,
+      "loss": 2.5173,
+      "step": 2999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361012771351766e-05,
+      "loss": 2.6271,
+      "step": 3000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4357362171671136e-05,
+      "loss": 2.4828,
+      "step": 3001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4353710855054238e-05,
+      "loss": 2.567,
+      "step": 3002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4350058822101827e-05,
+      "loss": 2.6298,
+      "step": 3003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43464060734148e-05,
+      "loss": 2.536,
+      "step": 3004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4342752609594155e-05,
+      "loss": 2.5368,
+      "step": 3005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4339098431241012e-05,
+      "loss": 2.4644,
+      "step": 3006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4335443538956613e-05,
+      "loss": 2.5413,
+      "step": 3007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433178793334231e-05,
+      "loss": 2.4867,
+      "step": 3008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4328131614999577e-05,
+      "loss": 2.4471,
+      "step": 3009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324474584530004e-05,
+      "loss": 2.5597,
+      "step": 3010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320816842535303e-05,
+      "loss": 2.5557,
+      "step": 3011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4317158389617291e-05,
+      "loss": 2.3443,
+      "step": 3012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4313499226377915e-05,
+      "loss": 2.5387,
+      "step": 3013
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4309839353419232e-05,
+      "loss": 2.533,
+      "step": 3014
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430617877134342e-05,
+      "loss": 2.5487,
+      "step": 3015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302517480752762e-05,
+      "loss": 2.5105,
+      "step": 3016
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429885548224968e-05,
+      "loss": 2.5705,
+      "step": 3017
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4295192776436688e-05,
+      "loss": 2.4856,
+      "step": 3018
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4291529363916431e-05,
+      "loss": 2.5621,
+      "step": 3019
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4287865245291668e-05,
+      "loss": 2.4777,
+      "step": 3020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4284200421165272e-05,
+      "loss": 2.5712,
+      "step": 3021
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280534892140231e-05,
+      "loss": 2.4203,
+      "step": 3022
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276868658819654e-05,
+      "loss": 2.5087,
+      "step": 3023
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427320172180676e-05,
+      "loss": 2.5517,
+      "step": 3024
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426953408170489e-05,
+      "loss": 2.4405,
+      "step": 3025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4265865739117493e-05,
+      "loss": 2.6176,
+      "step": 3026
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4262196694648138e-05,
+      "loss": 2.6042,
+      "step": 3027
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4258526948900513e-05,
+      "loss": 2.4701,
+      "step": 3028
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254856502478416e-05,
+      "loss": 2.444,
+      "step": 3029
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4251185355985761e-05,
+      "loss": 2.5675,
+      "step": 3030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247513510026579e-05,
+      "loss": 2.6129,
+      "step": 3031
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4243840965205016e-05,
+      "loss": 2.4831,
+      "step": 3032
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4240167722125326e-05,
+      "loss": 2.5386,
+      "step": 3033
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4236493781391896e-05,
+      "loss": 2.5151,
+      "step": 3034
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4232819143609204e-05,
+      "loss": 2.5524,
+      "step": 3035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229143809381862e-05,
+      "loss": 2.5567,
+      "step": 3036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4225467779314589e-05,
+      "loss": 2.5267,
+      "step": 3037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4221791054012215e-05,
+      "loss": 2.5713,
+      "step": 3038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4218113634079688e-05,
+      "loss": 2.5806,
+      "step": 3039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214435520122072e-05,
+      "loss": 2.5094,
+      "step": 3040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210756712744545e-05,
+      "loss": 2.5349,
+      "step": 3041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4207077212552398e-05,
+      "loss": 2.4985,
+      "step": 3042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4203397020151031e-05,
+      "loss": 2.3439,
+      "step": 3043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4199716136145965e-05,
+      "loss": 2.4634,
+      "step": 3044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196034561142833e-05,
+      "loss": 2.5604,
+      "step": 3045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419235229574738e-05,
+      "loss": 2.5086,
+      "step": 3046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4188669340565467e-05,
+      "loss": 2.4814,
+      "step": 3047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4184985696203065e-05,
+      "loss": 2.6387,
+      "step": 3048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181301363266261e-05,
+      "loss": 2.5231,
+      "step": 3049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417761634236126e-05,
+      "loss": 2.424,
+      "step": 3050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4173930634094369e-05,
+      "loss": 2.4749,
+      "step": 3051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4170244239072013e-05,
+      "loss": 2.5511,
+      "step": 3052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166557157900733e-05,
+      "loss": 2.5414,
+      "step": 3053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4162869391187184e-05,
+      "loss": 2.4594,
+      "step": 3054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4159180939538129e-05,
+      "loss": 2.6255,
+      "step": 3055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4155491803560439e-05,
+      "loss": 2.5981,
+      "step": 3056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4151801983861115e-05,
+      "loss": 2.5038,
+      "step": 3057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414811148104725e-05,
+      "loss": 2.5595,
+      "step": 3058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4144420295726069e-05,
+      "loss": 2.5508,
+      "step": 3059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4140728428504888e-05,
+      "loss": 2.4487,
+      "step": 3060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137035879991153e-05,
+      "loss": 2.4439,
+      "step": 3061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133342650792414e-05,
+      "loss": 2.6083,
+      "step": 3062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4129648741516338e-05,
+      "loss": 2.5503,
+      "step": 3063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412595415277069e-05,
+      "loss": 2.4704,
+      "step": 3064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412225888516337e-05,
+      "loss": 2.4393,
+      "step": 3065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118562939302372e-05,
+      "loss": 2.5607,
+      "step": 3066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4114866315795804e-05,
+      "loss": 2.5147,
+      "step": 3067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111169015251887e-05,
+      "loss": 2.4757,
+      "step": 3068
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4107471038278961e-05,
+      "loss": 2.4839,
+      "step": 3069
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4103772385485466e-05,
+      "loss": 2.4909,
+      "step": 3070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100073057479958e-05,
+      "loss": 2.4995,
+      "step": 3071
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096373054871105e-05,
+      "loss": 2.492,
+      "step": 3072
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092672378267684e-05,
+      "loss": 2.578,
+      "step": 3073
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4088971028278584e-05,
+      "loss": 2.4803,
+      "step": 3074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4085269005512805e-05,
+      "loss": 2.5818,
+      "step": 3075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408156631057946e-05,
+      "loss": 2.5051,
+      "step": 3076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4077862944087763e-05,
+      "loss": 2.4841,
+      "step": 3077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4074158906647053e-05,
+      "loss": 2.506,
+      "step": 3078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4070454198866767e-05,
+      "loss": 2.4672,
+      "step": 3079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4066748821356461e-05,
+      "loss": 2.4193,
+      "step": 3080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063042774725792e-05,
+      "loss": 2.5278,
+      "step": 3081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405933605958454e-05,
+      "loss": 2.5073,
+      "step": 3082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405562867654258e-05,
+      "loss": 2.4996,
+      "step": 3083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4051920626209908e-05,
+      "loss": 2.5593,
+      "step": 3084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4048211909196625e-05,
+      "loss": 2.4675,
+      "step": 3085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4044502526112944e-05,
+      "loss": 2.5102,
+      "step": 3086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4040792477569186e-05,
+      "loss": 2.4349,
+      "step": 3087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4037081764175777e-05,
+      "loss": 2.6063,
+      "step": 3088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033370386543268e-05,
+      "loss": 2.4736,
+      "step": 3089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4029658345282297e-05,
+      "loss": 2.6333,
+      "step": 3090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4025945641003632e-05,
+      "loss": 2.5313,
+      "step": 3091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4022232274318132e-05,
+      "loss": 2.6044,
+      "step": 3092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4018518245836784e-05,
+      "loss": 2.5141,
+      "step": 3093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4014803556170663e-05,
+      "loss": 2.5123,
+      "step": 3094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4011088205930971e-05,
+      "loss": 2.5401,
+      "step": 3095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400737219572901e-05,
+      "loss": 2.5097,
+      "step": 3096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003655526176192e-05,
+      "loss": 2.5498,
+      "step": 3097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399993819788403e-05,
+      "loss": 2.4917,
+      "step": 3098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3996220211464164e-05,
+      "loss": 2.521,
+      "step": 3099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992501567528326e-05,
+      "loss": 2.4955,
+      "step": 3100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3988782266688357e-05,
+      "loss": 2.4092,
+      "step": 3101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3985062309556214e-05,
+      "loss": 2.5818,
+      "step": 3102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3981341696743963e-05,
+      "loss": 2.5652,
+      "step": 3103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977620428863763e-05,
+      "loss": 2.5854,
+      "step": 3104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3973898506527896e-05,
+      "loss": 2.4961,
+      "step": 3105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3970175930348748e-05,
+      "loss": 2.454,
+      "step": 3106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3966452700938811e-05,
+      "loss": 2.5668,
+      "step": 3107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3962728818910682e-05,
+      "loss": 2.501,
+      "step": 3108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959004284877069e-05,
+      "loss": 2.4919,
+      "step": 3109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955279099450786e-05,
+      "loss": 2.5561,
+      "step": 3110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3951553263244752e-05,
+      "loss": 2.4558,
+      "step": 3111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3947826776872002e-05,
+      "loss": 2.557,
+      "step": 3112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3944099640945665e-05,
+      "loss": 2.5235,
+      "step": 3113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940371856078988e-05,
+      "loss": 2.4913,
+      "step": 3114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936643422885316e-05,
+      "loss": 2.5522,
+      "step": 3115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3932914341978112e-05,
+      "loss": 2.5632,
+      "step": 3116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3929184613970927e-05,
+      "loss": 2.4068,
+      "step": 3117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3925454239477439e-05,
+      "loss": 2.3992,
+      "step": 3118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3921723219111418e-05,
+      "loss": 2.4646,
+      "step": 3119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3917991553486752e-05,
+      "loss": 2.3956,
+      "step": 3120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3914259243217421e-05,
+      "loss": 2.4708,
+      "step": 3121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910526288917524e-05,
+      "loss": 2.5764,
+      "step": 3122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3906792691201254e-05,
+      "loss": 2.5145,
+      "step": 3123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903058450682927e-05,
+      "loss": 2.4553,
+      "step": 3124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899323567976944e-05,
+      "loss": 2.5147,
+      "step": 3125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389558804369783e-05,
+      "loss": 2.3781,
+      "step": 3126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891851878460202e-05,
+      "loss": 2.584,
+      "step": 3127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3888115072878787e-05,
+      "loss": 2.6339,
+      "step": 3128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884377627568426e-05,
+      "loss": 2.5447,
+      "step": 3129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388063954314405e-05,
+      "loss": 2.5035,
+      "step": 3130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3876900820220706e-05,
+      "loss": 2.5648,
+      "step": 3131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873161459413541e-05,
+      "loss": 2.4624,
+      "step": 3132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3869421461337813e-05,
+      "loss": 2.5074,
+      "step": 3133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3865680826608876e-05,
+      "loss": 2.4329,
+      "step": 3134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861939555842194e-05,
+      "loss": 2.5438,
+      "step": 3135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858197649653338e-05,
+      "loss": 2.5159,
+      "step": 3136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3854455108657979e-05,
+      "loss": 2.5967,
+      "step": 3137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850711933471892e-05,
+      "loss": 2.4506,
+      "step": 3138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846968124710959e-05,
+      "loss": 2.5475,
+      "step": 3139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3843223682991167e-05,
+      "loss": 2.4975,
+      "step": 3140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3839478608928608e-05,
+      "loss": 2.4862,
+      "step": 3141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835732903139467e-05,
+      "loss": 2.5191,
+      "step": 3142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831986566240052e-05,
+      "loss": 2.534,
+      "step": 3143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382823959884676e-05,
+      "loss": 2.4656,
+      "step": 3144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3824492001576096e-05,
+      "loss": 2.6696,
+      "step": 3145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3820743775044666e-05,
+      "loss": 2.4407,
+      "step": 3146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816994919869194e-05,
+      "loss": 2.5746,
+      "step": 3147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813245436666482e-05,
+      "loss": 2.5559,
+      "step": 3148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3809495326053454e-05,
+      "loss": 2.5702,
+      "step": 3149
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3805744588647138e-05,
+      "loss": 2.4738,
+      "step": 3150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3801993225064654e-05,
+      "loss": 2.4567,
+      "step": 3151
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379824123592323e-05,
+      "loss": 2.5328,
+      "step": 3152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3794488621840199e-05,
+      "loss": 2.5158,
+      "step": 3153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3790735383432998e-05,
+      "loss": 2.5546,
+      "step": 3154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378698152131916e-05,
+      "loss": 2.532,
+      "step": 3155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783227036116324e-05,
+      "loss": 2.489,
+      "step": 3156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3779471928442239e-05,
+      "loss": 2.5832,
+      "step": 3157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3775716198914744e-05,
+      "loss": 2.5351,
+      "step": 3158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771959848151784e-05,
+      "loss": 2.5266,
+      "step": 3159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3768202876771417e-05,
+      "loss": 2.5491,
+      "step": 3160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3764445285391785e-05,
+      "loss": 2.5241,
+      "step": 3161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3760687074631144e-05,
+      "loss": 2.5545,
+      "step": 3162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756928245107852e-05,
+      "loss": 2.5347,
+      "step": 3163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3753168797440366e-05,
+      "loss": 2.5879,
+      "step": 3164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3749408732247243e-05,
+      "loss": 2.5211,
+      "step": 3165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374564805014714e-05,
+      "loss": 2.4866,
+      "step": 3166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3741886751758827e-05,
+      "loss": 2.4795,
+      "step": 3167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3738124837701157e-05,
+      "loss": 2.4988,
+      "step": 3168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3734362308593101e-05,
+      "loss": 2.5976,
+      "step": 3169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3730599165053727e-05,
+      "loss": 2.3955,
+      "step": 3170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726835407702197e-05,
+      "loss": 2.5629,
+      "step": 3171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3723071037157778e-05,
+      "loss": 2.5174,
+      "step": 3172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719306054039842e-05,
+      "loss": 2.5849,
+      "step": 3173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3715540458967857e-05,
+      "loss": 2.5629,
+      "step": 3174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3711774252561398e-05,
+      "loss": 2.4192,
+      "step": 3175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708007435440127e-05,
+      "loss": 2.4874,
+      "step": 3176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704240008223822e-05,
+      "loss": 2.5055,
+      "step": 3177
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3700471971532351e-05,
+      "loss": 2.4793,
+      "step": 3178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696703325985687e-05,
+      "loss": 2.5466,
+      "step": 3179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3692934072203903e-05,
+      "loss": 2.564,
+      "step": 3180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3689164210807174e-05,
+      "loss": 2.5288,
+      "step": 3181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3685393742415764e-05,
+      "loss": 2.5733,
+      "step": 3182
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3681622667650053e-05,
+      "loss": 2.5439,
+      "step": 3183
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3677850987130506e-05,
+      "loss": 2.5082,
+      "step": 3184
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3674078701477703e-05,
+      "loss": 2.5638,
+      "step": 3185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670305811312305e-05,
+      "loss": 2.5934,
+      "step": 3186
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3666532317255087e-05,
+      "loss": 2.559,
+      "step": 3187
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3662758219926925e-05,
+      "loss": 2.4532,
+      "step": 3188
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3658983519948777e-05,
+      "loss": 2.4591,
+      "step": 3189
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365520821794172e-05,
+      "loss": 2.5489,
+      "step": 3190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651432314526915e-05,
+      "loss": 2.5435,
+      "step": 3191
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3647655810325634e-05,
+      "loss": 2.4565,
+      "step": 3192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3643878705959237e-05,
+      "loss": 2.4445,
+      "step": 3193
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3640101002049192e-05,
+      "loss": 2.5823,
+      "step": 3194
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3636322699217059e-05,
+      "loss": 2.5779,
+      "step": 3195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36325437980845e-05,
+      "loss": 2.6279,
+      "step": 3196
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3628764299273273e-05,
+      "loss": 2.4878,
+      "step": 3197
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362498420340524e-05,
+      "loss": 2.5345,
+      "step": 3198
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3621203511102353e-05,
+      "loss": 2.6499,
+      "step": 3199
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3617422222986668e-05,
+      "loss": 2.5429,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3613640339680335e-05,
+      "loss": 2.5135,
+      "step": 3201
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3609857861805607e-05,
+      "loss": 2.5019,
+      "step": 3202
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606074789984827e-05,
+      "loss": 2.5902,
+      "step": 3203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360229112484045e-05,
+      "loss": 2.5503,
+      "step": 3204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598506866995012e-05,
+      "loss": 2.4228,
+      "step": 3205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3594722017071153e-05,
+      "loss": 2.6243,
+      "step": 3206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3590936575691618e-05,
+      "loss": 2.488,
+      "step": 3207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3587150543479237e-05,
+      "loss": 2.5174,
+      "step": 3208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3583363921056945e-05,
+      "loss": 2.5429,
+      "step": 3209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3579576709047766e-05,
+      "loss": 2.5587,
+      "step": 3210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3575788908074838e-05,
+      "loss": 2.5523,
+      "step": 3211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357200051876138e-05,
+      "loss": 2.4544,
+      "step": 3212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3568211541730708e-05,
+      "loss": 2.5305,
+      "step": 3213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3564421977606241e-05,
+      "loss": 2.5042,
+      "step": 3214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35606318270115e-05,
+      "loss": 2.5196,
+      "step": 3215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3556841090570088e-05,
+      "loss": 2.495,
+      "step": 3216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3553049768905714e-05,
+      "loss": 2.5386,
+      "step": 3217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354925786264218e-05,
+      "loss": 2.5005,
+      "step": 3218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354546537240339e-05,
+      "loss": 2.4754,
+      "step": 3219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541672298813332e-05,
+      "loss": 2.5307,
+      "step": 3220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3537878642496103e-05,
+      "loss": 2.4682,
+      "step": 3221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353408440407589e-05,
+      "loss": 2.4754,
+      "step": 3222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3530289584176972e-05,
+      "loss": 2.4698,
+      "step": 3223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526494183423731e-05,
+      "loss": 2.5468,
+      "step": 3224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.352269820244064e-05,
+      "loss": 2.5199,
+      "step": 3225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3518901641852268e-05,
+      "loss": 2.589,
+      "step": 3226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515104502283283e-05,
+      "loss": 2.4681,
+      "step": 3227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511306784358444e-05,
+      "loss": 2.5077,
+      "step": 3228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3507508488702606e-05,
+      "loss": 2.5556,
+      "step": 3229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3503709615940718e-05,
+      "loss": 2.5173,
+      "step": 3230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499910166697827e-05,
+      "loss": 2.5019,
+      "step": 3231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3496110141599077e-05,
+      "loss": 2.5434,
+      "step": 3232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492309541269695e-05,
+      "loss": 2.4813,
+      "step": 3233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3488508366335017e-05,
+      "loss": 2.509,
+      "step": 3234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484706617420467e-05,
+      "loss": 2.4014,
+      "step": 3235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3480904295151562e-05,
+      "loss": 2.5536,
+      "step": 3236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3477101400153912e-05,
+      "loss": 2.4546,
+      "step": 3237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473297933053231e-05,
+      "loss": 2.4995,
+      "step": 3238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3469493894475316e-05,
+      "loss": 2.5119,
+      "step": 3239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3465689285046064e-05,
+      "loss": 2.5036,
+      "step": 3240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461884105391465e-05,
+      "loss": 2.5201,
+      "step": 3241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458078356137602e-05,
+      "loss": 2.6082,
+      "step": 3242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345427203791065e-05,
+      "loss": 2.4199,
+      "step": 3243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450465151336878e-05,
+      "loss": 2.4561,
+      "step": 3244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3446657697042657e-05,
+      "loss": 2.6004,
+      "step": 3245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344284967565444e-05,
+      "loss": 2.4556,
+      "step": 3246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3439041087798775e-05,
+      "loss": 2.4671,
+      "step": 3247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435231934102311e-05,
+      "loss": 2.4861,
+      "step": 3248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3431422215191785e-05,
+      "loss": 2.5332,
+      "step": 3249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3427611931694023e-05,
+      "loss": 2.5291,
+      "step": 3250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342380108423595e-05,
+      "loss": 2.4807,
+      "step": 3251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3419989673444582e-05,
+      "loss": 2.5633,
+      "step": 3252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3416177699947029e-05,
+      "loss": 2.4767,
+      "step": 3253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3412365164370487e-05,
+      "loss": 2.5483,
+      "step": 3254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3408552067342256e-05,
+      "loss": 2.5879,
+      "step": 3255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3404738409489718e-05,
+      "loss": 2.5278,
+      "step": 3256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340092419144035e-05,
+      "loss": 2.5042,
+      "step": 3257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3397109413821728e-05,
+      "loss": 2.5943,
+      "step": 3258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339329407726151e-05,
+      "loss": 2.4872,
+      "step": 3259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338947818238745e-05,
+      "loss": 2.5117,
+      "step": 3260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3385661729827395e-05,
+      "loss": 2.5247,
+      "step": 3261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381844720209286e-05,
+      "loss": 2.544,
+      "step": 3262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3378027154161149e-05,
+      "loss": 2.4991,
+      "step": 3263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374209032311103e-05,
+      "loss": 2.4991,
+      "step": 3264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370390355287368e-05,
+      "loss": 2.5047,
+      "step": 3265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3366571123718245e-05,
+      "loss": 2.4714,
+      "step": 3266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362751338232122e-05,
+      "loss": 2.4837,
+      "step": 3267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3358930999457496e-05,
+      "loss": 2.3829,
+      "step": 3268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335511010802294e-05,
+      "loss": 2.5223,
+      "step": 3269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3351288664557123e-05,
+      "loss": 2.5546,
+      "step": 3270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347466669688798e-05,
+      "loss": 2.5334,
+      "step": 3271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343644124046826e-05,
+      "loss": 2.5646,
+      "step": 3272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339821028260138e-05,
+      "loss": 2.5414,
+      "step": 3273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335997382957772e-05,
+      "loss": 2.555,
+      "step": 3274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3332173188768845e-05,
+      "loss": 2.5162,
+      "step": 3275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3328348446322572e-05,
+      "loss": 2.5601,
+      "step": 3276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332452315624825e-05,
+      "loss": 2.5147,
+      "step": 3277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3320697319175277e-05,
+      "loss": 2.5013,
+      "step": 3278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331687093573313e-05,
+      "loss": 2.5602,
+      "step": 3279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3313044006551386e-05,
+      "loss": 2.5512,
+      "step": 3280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3309216532259698e-05,
+      "loss": 2.4784,
+      "step": 3281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330538851348783e-05,
+      "loss": 2.4407,
+      "step": 3282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3301559950865615e-05,
+      "loss": 2.4258,
+      "step": 3283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3297730845022984e-05,
+      "loss": 2.6073,
+      "step": 3284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3293901196589958e-05,
+      "loss": 2.5084,
+      "step": 3285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3290071006196645e-05,
+      "loss": 2.5699,
+      "step": 3286
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3286240274473248e-05,
+      "loss": 2.5039,
+      "step": 3287
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3282409002050045e-05,
+      "loss": 2.512,
+      "step": 3288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3278577189557424e-05,
+      "loss": 2.6644,
+      "step": 3289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3274744837625842e-05,
+      "loss": 2.4837,
+      "step": 3290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270911946885855e-05,
+      "loss": 2.4432,
+      "step": 3291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3267078517968106e-05,
+      "loss": 2.4495,
+      "step": 3292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3263244551503327e-05,
+      "loss": 2.4715,
+      "step": 3293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259410048122337e-05,
+      "loss": 2.5532,
+      "step": 3294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3255575008456042e-05,
+      "loss": 2.4889,
+      "step": 3295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251739433135441e-05,
+      "loss": 2.5127,
+      "step": 3296
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3247903322791622e-05,
+      "loss": 2.5072,
+      "step": 3297
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3244066678055745e-05,
+      "loss": 2.6322,
+      "step": 3298
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3240229499559085e-05,
+      "loss": 2.4807,
+      "step": 3299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3236391787932978e-05,
+      "loss": 2.4184,
+      "step": 3300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3232553543808868e-05,
+      "loss": 2.4091,
+      "step": 3301
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228714767818274e-05,
+      "loss": 2.5875,
+      "step": 3302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3224875460592813e-05,
+      "loss": 2.4374,
+      "step": 3303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3221035622764176e-05,
+      "loss": 2.5648,
+      "step": 3304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217195254964148e-05,
+      "loss": 2.5556,
+      "step": 3305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3213354357824609e-05,
+      "loss": 2.3891,
+      "step": 3306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3209512931977515e-05,
+      "loss": 2.442,
+      "step": 3307
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3205670978054915e-05,
+      "loss": 2.5464,
+      "step": 3308
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3201828496688937e-05,
+      "loss": 2.4394,
+      "step": 3309
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319798548851181e-05,
+      "loss": 2.4493,
+      "step": 3310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3194141954155835e-05,
+      "loss": 2.5085,
+      "step": 3311
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3190297894253408e-05,
+      "loss": 2.4353,
+      "step": 3312
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318645330943701e-05,
+      "loss": 2.5184,
+      "step": 3313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3182608200339208e-05,
+      "loss": 2.508,
+      "step": 3314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317876256759265e-05,
+      "loss": 2.4968,
+      "step": 3315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3174916411830087e-05,
+      "loss": 2.5272,
+      "step": 3316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3171069733684332e-05,
+      "loss": 2.5283,
+      "step": 3317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31672225337883e-05,
+      "loss": 2.4526,
+      "step": 3318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316337481277499e-05,
+      "loss": 2.56,
+      "step": 3319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3159526571277483e-05,
+      "loss": 2.5282,
+      "step": 3320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3155677809928949e-05,
+      "loss": 2.5091,
+      "step": 3321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151828529362637e-05,
+      "loss": 2.4951,
+      "step": 3322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3147978730211896e-05,
+      "loss": 2.5208,
+      "step": 3323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314412841311014e-05,
+      "loss": 2.4727,
+      "step": 3324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140277578690882e-05,
+      "loss": 2.5612,
+      "step": 3325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3136426227587724e-05,
+      "loss": 2.5449,
+      "step": 3326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3132574360434338e-05,
+      "loss": 2.5917,
+      "step": 3327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3128721977864492e-05,
+      "loss": 2.505,
+      "step": 3328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3124869080512032e-05,
+      "loss": 2.4283,
+      "step": 3329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121015669010899e-05,
+      "loss": 2.4294,
+      "step": 3330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3117161743995108e-05,
+      "loss": 2.4556,
+      "step": 3331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3113307306098762e-05,
+      "loss": 2.532,
+      "step": 3332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310945235595605e-05,
+      "loss": 2.5339,
+      "step": 3333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105596894201247e-05,
+      "loss": 2.4965,
+      "step": 3334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3101740921468701e-05,
+      "loss": 2.4235,
+      "step": 3335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3097884438392862e-05,
+      "loss": 2.5225,
+      "step": 3336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3094027445608252e-05,
+      "loss": 2.4829,
+      "step": 3337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 2.3851,
+      "step": 3338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3086311933451228e-05,
+      "loss": 2.5514,
+      "step": 3339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308245341534828e-05,
+      "loss": 2.5587,
+      "step": 3340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.30785943900755e-05,
+      "loss": 2.507,
+      "step": 3341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3074734858267821e-05,
+      "loss": 2.5501,
+      "step": 3342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070874820560275e-05,
+      "loss": 2.5718,
+      "step": 3343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3067014277587971e-05,
+      "loss": 2.4853,
+      "step": 3344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3063153229986096e-05,
+      "loss": 2.5556,
+      "step": 3345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059291678389932e-05,
+      "loss": 2.4896,
+      "step": 3346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3055429623434834e-05,
+      "loss": 2.5462,
+      "step": 3347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3051567065756245e-05,
+      "loss": 2.4584,
+      "step": 3348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3047704005989685e-05,
+      "loss": 2.421,
+      "step": 3349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3043840444770766e-05,
+      "loss": 2.5643,
+      "step": 3350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3039976382735173e-05,
+      "loss": 2.5254,
+      "step": 3351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3036111820518677e-05,
+      "loss": 2.5654,
+      "step": 3352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3032246758757135e-05,
+      "loss": 2.4945,
+      "step": 3353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028381198086483e-05,
+      "loss": 2.5035,
+      "step": 3354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3024515139142736e-05,
+      "loss": 2.4308,
+      "step": 3355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3020648582561992e-05,
+      "loss": 2.5313,
+      "step": 3356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301678152898044e-05,
+      "loss": 2.5718,
+      "step": 3357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3012913979034339e-05,
+      "loss": 2.4999,
+      "step": 3358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3009045933360035e-05,
+      "loss": 2.475,
+      "step": 3359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3005177392593951e-05,
+      "loss": 2.5335,
+      "step": 3360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3001308357372601e-05,
+      "loss": 2.5724,
+      "step": 3361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2997438828332577e-05,
+      "loss": 2.5286,
+      "step": 3362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993568806110539e-05,
+      "loss": 2.6116,
+      "step": 3363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2989698291343248e-05,
+      "loss": 2.5444,
+      "step": 3364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2985827284667535e-05,
+      "loss": 2.4288,
+      "step": 3365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2981955786720312e-05,
+      "loss": 2.5793,
+      "step": 3366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297808379813858e-05,
+      "loss": 2.5462,
+      "step": 3367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974211319559407e-05,
+      "loss": 2.4979,
+      "step": 3368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297033835161995e-05,
+      "loss": 2.4051,
+      "step": 3369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2966464894957453e-05,
+      "loss": 2.536,
+      "step": 3370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962590950209229e-05,
+      "loss": 2.5383,
+      "step": 3371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295871651801267e-05,
+      "loss": 2.4897,
+      "step": 3372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954841599005258e-05,
+      "loss": 2.5762,
+      "step": 3373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2950966193824554e-05,
+      "loss": 2.6503,
+      "step": 3374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2947090303108193e-05,
+      "loss": 2.4647,
+      "step": 3375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294321392749389e-05,
+      "loss": 2.5136,
+      "step": 3376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2939337067619448e-05,
+      "loss": 2.4759,
+      "step": 3377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2935459724122739e-05,
+      "loss": 2.5433,
+      "step": 3378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2931581897641722e-05,
+      "loss": 2.4748,
+      "step": 3379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927703588814431e-05,
+      "loss": 2.551,
+      "step": 3380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2923824798278988e-05,
+      "loss": 2.5245,
+      "step": 3381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291994552667358e-05,
+      "loss": 2.5967,
+      "step": 3382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2916065774636481e-05,
+      "loss": 2.5656,
+      "step": 3383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291218554280605e-05,
+      "loss": 2.5059,
+      "step": 3384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2908304831820716e-05,
+      "loss": 2.5064,
+      "step": 3385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2904423642318986e-05,
+      "loss": 2.4853,
+      "step": 3386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900541974939456e-05,
+      "loss": 2.4781,
+      "step": 3387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2896659830320792e-05,
+      "loss": 2.4945,
+      "step": 3388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2892777209101737e-05,
+      "loss": 2.5016,
+      "step": 3389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2888894111921118e-05,
+      "loss": 2.4567,
+      "step": 3390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2885010539417842e-05,
+      "loss": 2.4447,
+      "step": 3391
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2881126492230887e-05,
+      "loss": 2.5378,
+      "step": 3392
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2877241970999307e-05,
+      "loss": 2.4832,
+      "step": 3393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2873356976362253e-05,
+      "loss": 2.4928,
+      "step": 3394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2869471508958928e-05,
+      "loss": 2.4843,
+      "step": 3395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2865585569428634e-05,
+      "loss": 2.4778,
+      "step": 3396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2861699158410736e-05,
+      "loss": 2.585,
+      "step": 3397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285781227654469e-05,
+      "loss": 2.4645,
+      "step": 3398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2853924924470014e-05,
+      "loss": 2.5219,
+      "step": 3399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2850037102826312e-05,
+      "loss": 2.4384,
+      "step": 3400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846148812253271e-05,
+      "loss": 2.5893,
+      "step": 3401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2842260053390644e-05,
+      "loss": 2.4804,
+      "step": 3402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2838370826878266e-05,
+      "loss": 2.4543,
+      "step": 3403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2834481133356051e-05,
+      "loss": 2.5756,
+      "step": 3404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2830590973463988e-05,
+      "loss": 2.5016,
+      "step": 3405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2826700347842139e-05,
+      "loss": 2.5038,
+      "step": 3406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822809257130646e-05,
+      "loss": 2.4669,
+      "step": 3407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281891770196973e-05,
+      "loss": 2.5039,
+      "step": 3408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281502568299969e-05,
+      "loss": 2.5356,
+      "step": 3409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2811133200860886e-05,
+      "loss": 2.5557,
+      "step": 3410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2807240256193776e-05,
+      "loss": 2.4352,
+      "step": 3411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2803346849638879e-05,
+      "loss": 2.4918,
+      "step": 3412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799452981836794e-05,
+      "loss": 2.4684,
+      "step": 3413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795558653428199e-05,
+      "loss": 2.4939,
+      "step": 3414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2791663865053848e-05,
+      "loss": 2.5098,
+      "step": 3415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2787768617354557e-05,
+      "loss": 2.5304,
+      "step": 3416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2783872910971238e-05,
+      "loss": 2.502,
+      "step": 3417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2779976746544867e-05,
+      "loss": 2.5027,
+      "step": 3418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2776080124716499e-05,
+      "loss": 2.602,
+      "step": 3419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2772183046127257e-05,
+      "loss": 2.481,
+      "step": 3420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276828551141835e-05,
+      "loss": 2.5261,
+      "step": 3421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2764387521231054e-05,
+      "loss": 2.4765,
+      "step": 3422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2760489076206723e-05,
+      "loss": 2.4991,
+      "step": 3423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2756590176986787e-05,
+      "loss": 2.5512,
+      "step": 3424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275269082421275e-05,
+      "loss": 2.4132,
+      "step": 3425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274879101852619e-05,
+      "loss": 2.4964,
+      "step": 3426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274489076056875e-05,
+      "loss": 2.4239,
+      "step": 3427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2740990050982174e-05,
+      "loss": 2.5476,
+      "step": 3428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2737088890408252e-05,
+      "loss": 2.3715,
+      "step": 3429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273318727948886e-05,
+      "loss": 2.4971,
+      "step": 3430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2729285218865947e-05,
+      "loss": 2.5061,
+      "step": 3431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2725382709181541e-05,
+      "loss": 2.5736,
+      "step": 3432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2721479751077737e-05,
+      "loss": 2.497,
+      "step": 3433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717576345196703e-05,
+      "loss": 2.5106,
+      "step": 3434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2713672492180688e-05,
+      "loss": 2.5189,
+      "step": 3435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709768192672014e-05,
+      "loss": 2.5578,
+      "step": 3436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2705863447313063e-05,
+      "loss": 2.5235,
+      "step": 3437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701958256746305e-05,
+      "loss": 2.4968,
+      "step": 3438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2698052621614281e-05,
+      "loss": 2.5269,
+      "step": 3439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26941465425596e-05,
+      "loss": 2.5154,
+      "step": 3440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2690240020224945e-05,
+      "loss": 2.4528,
+      "step": 3441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2686333055253075e-05,
+      "loss": 2.4608,
+      "step": 3442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2682425648286825e-05,
+      "loss": 2.4765,
+      "step": 3443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2678517799969088e-05,
+      "loss": 2.5185,
+      "step": 3444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267460951094285e-05,
+      "loss": 2.4619,
+      "step": 3445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2670700781851155e-05,
+      "loss": 2.5579,
+      "step": 3446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2666791613337122e-05,
+      "loss": 2.5784,
+      "step": 3447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2662882006043945e-05,
+      "loss": 2.4741,
+      "step": 3448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2658971960614894e-05,
+      "loss": 2.443,
+      "step": 3449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655061477693297e-05,
+      "loss": 2.5026,
+      "step": 3450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265115055792257e-05,
+      "loss": 2.5663,
+      "step": 3451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2647239201946193e-05,
+      "loss": 2.5197,
+      "step": 3452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2643327410407723e-05,
+      "loss": 2.5466,
+      "step": 3453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2639415183950771e-05,
+      "loss": 2.5059,
+      "step": 3454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2635502523219048e-05,
+      "loss": 2.5843,
+      "step": 3455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2631589428856313e-05,
+      "loss": 2.5608,
+      "step": 3456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627675901506408e-05,
+      "loss": 2.584,
+      "step": 3457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2623761941813241e-05,
+      "loss": 2.5052,
+      "step": 3458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2619847550420799e-05,
+      "loss": 2.5739,
+      "step": 3459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2615932727973129e-05,
+      "loss": 2.5335,
+      "step": 3460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2612017475114352e-05,
+      "loss": 2.5187,
+      "step": 3461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2608101792488667e-05,
+      "loss": 2.507,
+      "step": 3462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2604185680740337e-05,
+      "loss": 2.5099,
+      "step": 3463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26002691405137e-05,
+      "loss": 2.5469,
+      "step": 3464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2596352172453154e-05,
+      "loss": 2.5276,
+      "step": 3465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2592434777203186e-05,
+      "loss": 2.556,
+      "step": 3466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2588516955408336e-05,
+      "loss": 2.5354,
+      "step": 3467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258459870771322e-05,
+      "loss": 2.4712,
+      "step": 3468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2580680034762529e-05,
+      "loss": 2.577,
+      "step": 3469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2576760937201018e-05,
+      "loss": 2.5628,
+      "step": 3470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257284141567351e-05,
+      "loss": 2.479,
+      "step": 3471
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2568921470824909e-05,
+      "loss": 2.5322,
+      "step": 3472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565001103300171e-05,
+      "loss": 2.4065,
+      "step": 3473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2561080313744345e-05,
+      "loss": 2.5065,
+      "step": 3474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255715910280252e-05,
+      "loss": 2.5879,
+      "step": 3475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553237471119883e-05,
+      "loss": 2.5528,
+      "step": 3476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2549315419341675e-05,
+      "loss": 2.5047,
+      "step": 3477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2545392948113202e-05,
+      "loss": 2.5963,
+      "step": 3478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2541470058079855e-05,
+      "loss": 2.5615,
+      "step": 3479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537546749887078e-05,
+      "loss": 2.5886,
+      "step": 3480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2533623024180393e-05,
+      "loss": 2.4928,
+      "step": 3481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529698881605387e-05,
+      "loss": 2.5035,
+      "step": 3482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252577432280772e-05,
+      "loss": 2.5203,
+      "step": 3483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521849348433113e-05,
+      "loss": 2.6415,
+      "step": 3484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251792395912736e-05,
+      "loss": 2.3773,
+      "step": 3485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2513998155536324e-05,
+      "loss": 2.5503,
+      "step": 3486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251007193830594e-05,
+      "loss": 2.6036,
+      "step": 3487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2506145308082195e-05,
+      "loss": 2.4762,
+      "step": 3488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502218265511166e-05,
+      "loss": 2.4902,
+      "step": 3489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249829081123898e-05,
+      "loss": 2.4888,
+      "step": 3490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2494362945911843e-05,
+      "loss": 2.5213,
+      "step": 3491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2490434670176016e-05,
+      "loss": 2.3649,
+      "step": 3492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486505984677848e-05,
+      "loss": 2.5045,
+      "step": 3493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2482576890063736e-05,
+      "loss": 2.4933,
+      "step": 3494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247864738698015e-05,
+      "loss": 2.5831,
+      "step": 3495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2474717476073632e-05,
+      "loss": 2.5321,
+      "step": 3496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247078715799079e-05,
+      "loss": 2.5281,
+      "step": 3497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2466856433378289e-05,
+      "loss": 2.5019,
+      "step": 3498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2462925302882875e-05,
+      "loss": 2.4176,
+      "step": 3499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458993767151355e-05,
+      "loss": 2.5192,
+      "step": 3500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2455061826830599e-05,
+      "loss": 2.4619,
+      "step": 3501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2451129482567546e-05,
+      "loss": 2.5424,
+      "step": 3502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2447196735009204e-05,
+      "loss": 2.4649,
+      "step": 3503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244326358480265e-05,
+      "loss": 2.5446,
+      "step": 3504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2439330032595012e-05,
+      "loss": 2.5472,
+      "step": 3505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2435396079033505e-05,
+      "loss": 2.5415,
+      "step": 3506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2431461724765394e-05,
+      "loss": 2.5048,
+      "step": 3507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242752697043802e-05,
+      "loss": 2.4828,
+      "step": 3508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2423591816698781e-05,
+      "loss": 2.4811,
+      "step": 3509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2419656264195152e-05,
+      "loss": 2.6112,
+      "step": 3510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2415720313574662e-05,
+      "loss": 2.5114,
+      "step": 3511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241178396548491e-05,
+      "loss": 2.4527,
+      "step": 3512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2407847220573563e-05,
+      "loss": 2.5525,
+      "step": 3513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2403910079488352e-05,
+      "loss": 2.4893,
+      "step": 3514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2399972542877072e-05,
+      "loss": 2.4798,
+      "step": 3515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2396034611387581e-05,
+      "loss": 2.5032,
+      "step": 3516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392096285667808e-05,
+      "loss": 2.5175,
+      "step": 3517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388157566365741e-05,
+      "loss": 2.5921,
+      "step": 3518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2384218454129435e-05,
+      "loss": 2.5305,
+      "step": 3519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2380278949607011e-05,
+      "loss": 2.5734,
+      "step": 3520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2376339053446656e-05,
+      "loss": 2.5429,
+      "step": 3521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237239876629661e-05,
+      "loss": 2.4549,
+      "step": 3522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2368458088805194e-05,
+      "loss": 2.568,
+      "step": 3523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2364517021620783e-05,
+      "loss": 2.4672,
+      "step": 3524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2360575565391819e-05,
+      "loss": 2.5523,
+      "step": 3525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356633720766804e-05,
+      "loss": 2.5886,
+      "step": 3526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2352691488394315e-05,
+      "loss": 2.6305,
+      "step": 3527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2348748868922974e-05,
+      "loss": 2.5405,
+      "step": 3528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2344805863001488e-05,
+      "loss": 2.4603,
+      "step": 3529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2340862471278612e-05,
+      "loss": 2.5202,
+      "step": 3530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2336918694403173e-05,
+      "loss": 2.5856,
+      "step": 3531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2332974533024053e-05,
+      "loss": 2.401,
+      "step": 3532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329029987790207e-05,
+      "loss": 2.5685,
+      "step": 3533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2325085059350647e-05,
+      "loss": 2.5564,
+      "step": 3534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2321139748354454e-05,
+      "loss": 2.5521,
+      "step": 3535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231719405545076e-05,
+      "loss": 2.5802,
+      "step": 3536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2313247981288772e-05,
+      "loss": 2.5799,
+      "step": 3537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2309301526517756e-05,
+      "loss": 2.5803,
+      "step": 3538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2305354691787038e-05,
+      "loss": 2.5141,
+      "step": 3539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2301407477746008e-05,
+      "loss": 2.5219,
+      "step": 3540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2297459885044118e-05,
+      "loss": 2.6029,
+      "step": 3541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2293511914330887e-05,
+      "loss": 2.4768,
+      "step": 3542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2289563566255888e-05,
+      "loss": 2.5054,
+      "step": 3543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2285614841468767e-05,
+      "loss": 2.5414,
+      "step": 3544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2281665740619218e-05,
+      "loss": 2.5519,
+      "step": 3545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2277716264357004e-05,
+      "loss": 2.5405,
+      "step": 3546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2273766413331956e-05,
+      "loss": 2.4777,
+      "step": 3547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2269816188193958e-05,
+      "loss": 2.5724,
+      "step": 3548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2265865589592953e-05,
+      "loss": 2.407,
+      "step": 3549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2261914618178958e-05,
+      "loss": 2.4878,
+      "step": 3550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257963274602043e-05,
+      "loss": 2.6173,
+      "step": 3551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2254011559512338e-05,
+      "loss": 2.5661,
+      "step": 3552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2250059473560032e-05,
+      "loss": 2.597,
+      "step": 3553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224610701739539e-05,
+      "loss": 2.4469,
+      "step": 3554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224215419166872e-05,
+      "loss": 2.5423,
+      "step": 3555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238200997030398e-05,
+      "loss": 2.4739,
+      "step": 3556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2234247434130864e-05,
+      "loss": 2.4513,
+      "step": 3557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2230293503620615e-05,
+      "loss": 2.5145,
+      "step": 3558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2226339206150205e-05,
+      "loss": 2.4603,
+      "step": 3559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222384542370254e-05,
+      "loss": 2.4679,
+      "step": 3560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2218429512931444e-05,
+      "loss": 2.5756,
+      "step": 3561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2214474118484509e-05,
+      "loss": 2.4092,
+      "step": 3562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2210518359680251e-05,
+      "loss": 2.4834,
+      "step": 3563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2206562237169525e-05,
+      "loss": 2.4905,
+      "step": 3564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2202605751603255e-05,
+      "loss": 2.4237,
+      "step": 3565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2198648903632416e-05,
+      "loss": 2.5229,
+      "step": 3566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194691693908043e-05,
+      "loss": 2.5167,
+      "step": 3567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2190734123081237e-05,
+      "loss": 2.5529,
+      "step": 3568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218677619180316e-05,
+      "loss": 2.5352,
+      "step": 3569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2182817900725015e-05,
+      "loss": 2.4089,
+      "step": 3570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2178859250498089e-05,
+      "loss": 2.4796,
+      "step": 3571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217490024177371e-05,
+      "loss": 2.5323,
+      "step": 3572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170940875203274e-05,
+      "loss": 2.4343,
+      "step": 3573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2166981151438232e-05,
+      "loss": 2.429,
+      "step": 3574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163021071130101e-05,
+      "loss": 2.4695,
+      "step": 3575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215906063493044e-05,
+      "loss": 2.5165,
+      "step": 3576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2155099843490885e-05,
+      "loss": 2.491,
+      "step": 3577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2151138697463121e-05,
+      "loss": 2.4877,
+      "step": 3578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147177197498891e-05,
+      "loss": 2.4767,
+      "step": 3579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214321534425e-05,
+      "loss": 2.4904,
+      "step": 3580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213925313836831e-05,
+      "loss": 2.4219,
+      "step": 3581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2135290580505741e-05,
+      "loss": 2.5314,
+      "step": 3582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2131327671314266e-05,
+      "loss": 2.4864,
+      "step": 3583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2127364411445925e-05,
+      "loss": 2.3853,
+      "step": 3584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2123400801552809e-05,
+      "loss": 2.593,
+      "step": 3585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211943684228707e-05,
+      "loss": 2.5747,
+      "step": 3586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115472534300912e-05,
+      "loss": 2.4615,
+      "step": 3587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2111507878246606e-05,
+      "loss": 2.5258,
+      "step": 3588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210754287477647e-05,
+      "loss": 2.4948,
+      "step": 3589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2103577524542884e-05,
+      "loss": 2.5253,
+      "step": 3590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099611828198286e-05,
+      "loss": 2.5932,
+      "step": 3591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2095645786395171e-05,
+      "loss": 2.5484,
+      "step": 3592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091679399786087e-05,
+      "loss": 2.5469,
+      "step": 3593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2087712669023643e-05,
+      "loss": 2.5307,
+      "step": 3594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2083745594760503e-05,
+      "loss": 2.567,
+      "step": 3595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2079778177649388e-05,
+      "loss": 2.4884,
+      "step": 3596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2075810418343075e-05,
+      "loss": 2.4295,
+      "step": 3597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2071842317494391e-05,
+      "loss": 2.5298,
+      "step": 3598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067873875756236e-05,
+      "loss": 2.4985,
+      "step": 3599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2063905093781545e-05,
+      "loss": 2.5405,
+      "step": 3600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205993597222333e-05,
+      "loss": 2.5241,
+      "step": 3601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055966511734636e-05,
+      "loss": 2.4932,
+      "step": 3602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051996712968585e-05,
+      "loss": 2.5272,
+      "step": 3603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2048026576578344e-05,
+      "loss": 2.479,
+      "step": 3604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204405610321714e-05,
+      "loss": 2.5402,
+      "step": 3605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2040085293538244e-05,
+      "loss": 2.5536,
+      "step": 3606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2036114148194999e-05,
+      "loss": 2.5006,
+      "step": 3607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032142667840794e-05,
+      "loss": 2.5934,
+      "step": 3608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2028170853129073e-05,
+      "loss": 2.4122,
+      "step": 3609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2024198704713335e-05,
+      "loss": 2.5825,
+      "step": 3610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2020226223247136e-05,
+      "loss": 2.5136,
+      "step": 3611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2016253409384088e-05,
+      "loss": 2.5601,
+      "step": 3612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012280263777858e-05,
+      "loss": 2.5663,
+      "step": 3613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2008306787082157e-05,
+      "loss": 2.5045,
+      "step": 3614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200433297995077e-05,
+      "loss": 2.5188,
+      "step": 3615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2000358843037518e-05,
+      "loss": 2.523,
+      "step": 3616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199638437699628e-05,
+      "loss": 2.5545,
+      "step": 3617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1992409582481004e-05,
+      "loss": 2.4728,
+      "step": 3618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198843446014567e-05,
+      "loss": 2.5077,
+      "step": 3619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1984459010644329e-05,
+      "loss": 2.5805,
+      "step": 3620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1980483234631078e-05,
+      "loss": 2.48,
+      "step": 3621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1976507132760067e-05,
+      "loss": 2.5114,
+      "step": 3622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1972530705685506e-05,
+      "loss": 2.4478,
+      "step": 3623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1968553954061648e-05,
+      "loss": 2.4817,
+      "step": 3624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196457687854281e-05,
+      "loss": 2.4979,
+      "step": 3625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1960599479783362e-05,
+      "loss": 2.4558,
+      "step": 3626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1956621758437715e-05,
+      "loss": 2.4853,
+      "step": 3627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952643715160343e-05,
+      "loss": 2.4445,
+      "step": 3628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1948665350605777e-05,
+      "loss": 2.4999,
+      "step": 3629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1944686665428593e-05,
+      "loss": 2.4632,
+      "step": 3630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1940707660283412e-05,
+      "loss": 2.4827,
+      "step": 3631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1936728335824933e-05,
+      "loss": 2.5407,
+      "step": 3632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1932748692707883e-05,
+      "loss": 2.4562,
+      "step": 3633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1928768731587053e-05,
+      "loss": 2.5132,
+      "step": 3634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1924788453117282e-05,
+      "loss": 2.5624,
+      "step": 3635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1920807857953465e-05,
+      "loss": 2.5809,
+      "step": 3636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1916826946750543e-05,
+      "loss": 2.4631,
+      "step": 3637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1912845720163518e-05,
+      "loss": 2.5629,
+      "step": 3638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190886417884744e-05,
+      "loss": 2.5422,
+      "step": 3639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1904882323457403e-05,
+      "loss": 2.5136,
+      "step": 3640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1900900154648565e-05,
+      "loss": 2.4922,
+      "step": 3641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896917673076127e-05,
+      "loss": 2.5624,
+      "step": 3642
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892934879395351e-05,
+      "loss": 2.6316,
+      "step": 3643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1888951774261535e-05,
+      "loss": 2.4964,
+      "step": 3644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188496835833004e-05,
+      "loss": 2.5046,
+      "step": 3645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1880984632256276e-05,
+      "loss": 2.5076,
+      "step": 3646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1877000596695707e-05,
+      "loss": 2.5314,
+      "step": 3647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1873016252303836e-05,
+      "loss": 2.5816,
+      "step": 3648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1869031599736234e-05,
+      "loss": 2.5717,
+      "step": 3649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1865046639648506e-05,
+      "loss": 2.5422,
+      "step": 3650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1861061372696319e-05,
+      "loss": 2.4659,
+      "step": 3651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857075799535387e-05,
+      "loss": 2.5247,
+      "step": 3652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1853089920821474e-05,
+      "loss": 2.4603,
+      "step": 3653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1849103737210394e-05,
+      "loss": 2.5419,
+      "step": 3654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1845117249358008e-05,
+      "loss": 2.4542,
+      "step": 3655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1841130457920241e-05,
+      "loss": 2.5237,
+      "step": 3656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1837143363553048e-05,
+      "loss": 2.5027,
+      "step": 3657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1833155966912444e-05,
+      "loss": 2.5254,
+      "step": 3658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1829168268654496e-05,
+      "loss": 2.5042,
+      "step": 3659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1825180269435321e-05,
+      "loss": 2.5196,
+      "step": 3660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1821191969911075e-05,
+      "loss": 2.5048,
+      "step": 3661
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1817203370737975e-05,
+      "loss": 2.5967,
+      "step": 3662
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1813214472572282e-05,
+      "loss": 2.4475,
+      "step": 3663
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1809225276070312e-05,
+      "loss": 2.4905,
+      "step": 3664
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805235781888413e-05,
+      "loss": 2.4962,
+      "step": 3665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1801245990683008e-05,
+      "loss": 2.489,
+      "step": 3666
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797255903110547e-05,
+      "loss": 2.5317,
+      "step": 3667
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1793265519827542e-05,
+      "loss": 2.4661,
+      "step": 3668
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1789274841490543e-05,
+      "loss": 2.3958,
+      "step": 3669
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1785283868756162e-05,
+      "loss": 2.4702,
+      "step": 3670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1781292602281042e-05,
+      "loss": 2.4526,
+      "step": 3671
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1777301042721892e-05,
+      "loss": 2.5368,
+      "step": 3672
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1773309190735454e-05,
+      "loss": 2.3794,
+      "step": 3673
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1769317046978532e-05,
+      "loss": 2.502,
+      "step": 3674
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176532461210797e-05,
+      "loss": 2.4776,
+      "step": 3675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1761331886780656e-05,
+      "loss": 2.6072,
+      "step": 3676
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1757338871653536e-05,
+      "loss": 2.5815,
+      "step": 3677
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1753345567383595e-05,
+      "loss": 2.4344,
+      "step": 3678
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749351974627874e-05,
+      "loss": 2.4366,
+      "step": 3679
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174535809404345e-05,
+      "loss": 2.5962,
+      "step": 3680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741363926287463e-05,
+      "loss": 2.4841,
+      "step": 3681
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1737369472017081e-05,
+      "loss": 2.5443,
+      "step": 3682
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733374731889539e-05,
+      "loss": 2.5014,
+      "step": 3683
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1729379706562104e-05,
+      "loss": 2.5083,
+      "step": 3684
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725384396692095e-05,
+      "loss": 2.529,
+      "step": 3685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172138880293688e-05,
+      "loss": 2.4926,
+      "step": 3686
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1717392925953874e-05,
+      "loss": 2.4998,
+      "step": 3687
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1713396766400531e-05,
+      "loss": 2.5728,
+      "step": 3688
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1709400324934361e-05,
+      "loss": 2.4462,
+      "step": 3689
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1705403602212915e-05,
+      "loss": 2.527,
+      "step": 3690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1701406598893794e-05,
+      "loss": 2.4176,
+      "step": 3691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1697409315634638e-05,
+      "loss": 2.4553,
+      "step": 3692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1693411753093142e-05,
+      "loss": 2.4434,
+      "step": 3693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1689413911927044e-05,
+      "loss": 2.5389,
+      "step": 3694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685415792794118e-05,
+      "loss": 2.4944,
+      "step": 3695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1681417396352206e-05,
+      "loss": 2.4503,
+      "step": 3696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167741872325917e-05,
+      "loss": 2.502,
+      "step": 3697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1673419774172936e-05,
+      "loss": 2.4925,
+      "step": 3698
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669420549751464e-05,
+      "loss": 2.4657,
+      "step": 3699
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1665421050652774e-05,
+      "loss": 2.4711,
+      "step": 3700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166142127753491e-05,
+      "loss": 2.5281,
+      "step": 3701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1657421231055979e-05,
+      "loss": 2.5485,
+      "step": 3702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1653420911874123e-05,
+      "loss": 2.4754,
+      "step": 3703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1649420320647538e-05,
+      "loss": 2.4434,
+      "step": 3704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1645419458034453e-05,
+      "loss": 2.5635,
+      "step": 3705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1641418324693152e-05,
+      "loss": 2.5009,
+      "step": 3706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1637416921281957e-05,
+      "loss": 2.4463,
+      "step": 3707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1633415248459236e-05,
+      "loss": 2.5154,
+      "step": 3708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1629413306883404e-05,
+      "loss": 2.4519,
+      "step": 3709
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162541109721292e-05,
+      "loss": 2.5196,
+      "step": 3710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1621408620106282e-05,
+      "loss": 2.4753,
+      "step": 3711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1617405876222037e-05,
+      "loss": 2.5637,
+      "step": 3712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1613402866218777e-05,
+      "loss": 2.5331,
+      "step": 3713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609399590755128e-05,
+      "loss": 2.4977,
+      "step": 3714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1605396050489772e-05,
+      "loss": 2.5749,
+      "step": 3715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1601392246081428e-05,
+      "loss": 2.4345,
+      "step": 3716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597388178188865e-05,
+      "loss": 2.5589,
+      "step": 3717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593383847470882e-05,
+      "loss": 2.6176,
+      "step": 3718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1589379254586332e-05,
+      "loss": 2.5306,
+      "step": 3719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158537440019411e-05,
+      "loss": 2.4934,
+      "step": 3720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1581369284953157e-05,
+      "loss": 2.5243,
+      "step": 3721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577363909522441e-05,
+      "loss": 2.5432,
+      "step": 3722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573358274560993e-05,
+      "loss": 2.3811,
+      "step": 3723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569352380727876e-05,
+      "loss": 2.4566,
+      "step": 3724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1565346228682199e-05,
+      "loss": 2.5699,
+      "step": 3725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1561339819083109e-05,
+      "loss": 2.5405,
+      "step": 3726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1557333152589803e-05,
+      "loss": 2.5979,
+      "step": 3727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155332622986151e-05,
+      "loss": 2.529,
+      "step": 3728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154931905155751e-05,
+      "loss": 2.42,
+      "step": 3729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154531161833712e-05,
+      "loss": 2.4505,
+      "step": 3730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154130393085971e-05,
+      "loss": 2.5911,
+      "step": 3731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1537295989784669e-05,
+      "loss": 2.4534,
+      "step": 3732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1533287795771448e-05,
+      "loss": 2.4883,
+      "step": 3733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1529279349479537e-05,
+      "loss": 2.4964,
+      "step": 3734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1525270651568457e-05,
+      "loss": 2.444,
+      "step": 3735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1521261702697777e-05,
+      "loss": 2.4759,
+      "step": 3736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1517252503527112e-05,
+      "loss": 2.5132,
+      "step": 3737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1513243054716114e-05,
+      "loss": 2.5643,
+      "step": 3738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150923335692447e-05,
+      "loss": 2.5068,
+      "step": 3739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1505223410811914e-05,
+      "loss": 2.5342,
+      "step": 3740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1501213217038225e-05,
+      "loss": 2.4452,
+      "step": 3741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497202776263219e-05,
+      "loss": 2.5273,
+      "step": 3742
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1493192089146743e-05,
+      "loss": 2.4922,
+      "step": 3743
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14891811563487e-05,
+      "loss": 2.4781,
+      "step": 3744
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1485169978529027e-05,
+      "loss": 2.5046,
+      "step": 3745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14811585563477e-05,
+      "loss": 2.5054,
+      "step": 3746
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1477146890464734e-05,
+      "loss": 2.4686,
+      "step": 3747
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147313498154019e-05,
+      "loss": 2.5779,
+      "step": 3748
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1469122830234163e-05,
+      "loss": 2.4922,
+      "step": 3749
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1465110437206788e-05,
+      "loss": 2.5422,
+      "step": 3750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1461097803118246e-05,
+      "loss": 2.4518,
+      "step": 3751
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1457084928628755e-05,
+      "loss": 2.4764,
+      "step": 3752
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1453071814398565e-05,
+      "loss": 2.4542,
+      "step": 3753
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449058461087974e-05,
+      "loss": 2.5377,
+      "step": 3754
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144504486935732e-05,
+      "loss": 2.4796,
+      "step": 3755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441031039866974e-05,
+      "loss": 2.4732,
+      "step": 3756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1437016973277351e-05,
+      "loss": 2.463,
+      "step": 3757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14330026702489e-05,
+      "loss": 2.6066,
+      "step": 3758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1428988131442118e-05,
+      "loss": 2.5654,
+      "step": 3759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1424973357517531e-05,
+      "loss": 2.5674,
+      "step": 3760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1420958349135708e-05,
+      "loss": 2.5008,
+      "step": 3761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1416943106957258e-05,
+      "loss": 2.4856,
+      "step": 3762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1412927631642826e-05,
+      "loss": 2.483,
+      "step": 3763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408911923853093e-05,
+      "loss": 2.5034,
+      "step": 3764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404895984248788e-05,
+      "loss": 2.5297,
+      "step": 3765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1400879813490663e-05,
+      "loss": 2.4856,
+      "step": 3766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1396863412239522e-05,
+      "loss": 2.5544,
+      "step": 3767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392846781156201e-05,
+      "loss": 2.5377,
+      "step": 3768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1388829920901579e-05,
+      "loss": 2.5762,
+      "step": 3769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1384812832136554e-05,
+      "loss": 2.5426,
+      "step": 3770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1380795515522091e-05,
+      "loss": 2.4636,
+      "step": 3771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137677797171917e-05,
+      "loss": 2.5374,
+      "step": 3772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137276020138881e-05,
+      "loss": 2.5391,
+      "step": 3773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1368742205192085e-05,
+      "loss": 2.4643,
+      "step": 3774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1364723983790085e-05,
+      "loss": 2.4891,
+      "step": 3775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136070553784395e-05,
+      "loss": 2.4055,
+      "step": 3776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1356686868014845e-05,
+      "loss": 2.4963,
+      "step": 3777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1352667974963993e-05,
+      "loss": 2.5233,
+      "step": 3778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1348648859352631e-05,
+      "loss": 2.4982,
+      "step": 3779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1344629521842043e-05,
+      "loss": 2.5648,
+      "step": 3780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134060996309355e-05,
+      "loss": 2.4483,
+      "step": 3781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133659018376851e-05,
+      "loss": 2.4875,
+      "step": 3782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1332570184528313e-05,
+      "loss": 2.5294,
+      "step": 3783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1328549966034386e-05,
+      "loss": 2.5294,
+      "step": 3784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324529528948193e-05,
+      "loss": 2.4765,
+      "step": 3785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320508873931244e-05,
+      "loss": 2.5345,
+      "step": 3786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1316488001645061e-05,
+      "loss": 2.5267,
+      "step": 3787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1312466912751226e-05,
+      "loss": 2.525,
+      "step": 3788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1308445607911343e-05,
+      "loss": 2.5858,
+      "step": 3789
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1304424087787056e-05,
+      "loss": 2.5851,
+      "step": 3790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1300402353040046e-05,
+      "loss": 2.4418,
+      "step": 3791
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1296380404332022e-05,
+      "loss": 2.5225,
+      "step": 3792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.129235824232474e-05,
+      "loss": 2.4683,
+      "step": 3793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128833586767997e-05,
+      "loss": 2.649,
+      "step": 3794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1284313281059554e-05,
+      "loss": 2.447,
+      "step": 3795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1280290483125327e-05,
+      "loss": 2.4764,
+      "step": 3796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1276267474539184e-05,
+      "loss": 2.5328,
+      "step": 3797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1272244255963049e-05,
+      "loss": 2.4572,
+      "step": 3798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1268220828058884e-05,
+      "loss": 2.4818,
+      "step": 3799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1264197191488677e-05,
+      "loss": 2.5127,
+      "step": 3800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260173346914453e-05,
+      "loss": 2.5109,
+      "step": 3801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1256149294998278e-05,
+      "loss": 2.4313,
+      "step": 3802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252125036402246e-05,
+      "loss": 2.5228,
+      "step": 3803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1248100571788481e-05,
+      "loss": 2.5076,
+      "step": 3804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244075901819156e-05,
+      "loss": 2.5061,
+      "step": 3805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.124005102715646e-05,
+      "loss": 2.4146,
+      "step": 3806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1236025948462626e-05,
+      "loss": 2.4529,
+      "step": 3807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1232000666399919e-05,
+      "loss": 2.6015,
+      "step": 3808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1227975181630635e-05,
+      "loss": 2.5274,
+      "step": 3809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1223949494817105e-05,
+      "loss": 2.5424,
+      "step": 3810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1219923606621697e-05,
+      "loss": 2.5108,
+      "step": 3811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1215897517706806e-05,
+      "loss": 2.5344,
+      "step": 3812
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1211871228734858e-05,
+      "loss": 2.4902,
+      "step": 3813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1207844740368323e-05,
+      "loss": 2.4615,
+      "step": 3814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1203818053269692e-05,
+      "loss": 2.4661,
+      "step": 3815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1199791168101501e-05,
+      "loss": 2.5777,
+      "step": 3816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1195764085526304e-05,
+      "loss": 2.4526,
+      "step": 3817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1191736806206696e-05,
+      "loss": 2.5374,
+      "step": 3818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1187709330805305e-05,
+      "loss": 2.5569,
+      "step": 3819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1183681659984793e-05,
+      "loss": 2.5433,
+      "step": 3820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179653794407841e-05,
+      "loss": 2.5213,
+      "step": 3821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175625734737184e-05,
+      "loss": 2.531,
+      "step": 3822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1171597481635568e-05,
+      "loss": 2.529,
+      "step": 3823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116756903576578e-05,
+      "loss": 2.5194,
+      "step": 3824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116354039779064e-05,
+      "loss": 2.5152,
+      "step": 3825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1159511568373003e-05,
+      "loss": 2.5709,
+      "step": 3826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115548254817574e-05,
+      "loss": 2.5623,
+      "step": 3827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115145333786177e-05,
+      "loss": 2.4844,
+      "step": 3828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147423938094038e-05,
+      "loss": 2.585,
+      "step": 3829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1143394349535518e-05,
+      "loss": 2.4567,
+      "step": 3830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139364572849211e-05,
+      "loss": 2.4701,
+      "step": 3831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113533460869816e-05,
+      "loss": 2.4935,
+      "step": 3832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1131304457745433e-05,
+      "loss": 2.4956,
+      "step": 3833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127274120654126e-05,
+      "loss": 2.5433,
+      "step": 3834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112324359808737e-05,
+      "loss": 2.4225,
+      "step": 3835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119212890708324e-05,
+      "loss": 2.5554,
+      "step": 3836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111518199918018e-05,
+      "loss": 2.5351,
+      "step": 3837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1111150924166154e-05,
+      "loss": 2.4418,
+      "step": 3838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1107119666329504e-05,
+      "loss": 2.5627,
+      "step": 3839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1103088226333505e-05,
+      "loss": 2.5367,
+      "step": 3840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1099056604841469e-05,
+      "loss": 2.6014,
+      "step": 3841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1095024802516738e-05,
+      "loss": 2.5403,
+      "step": 3842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1090992820022685e-05,
+      "loss": 2.5788,
+      "step": 3843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086960658022705e-05,
+      "loss": 2.4377,
+      "step": 3844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108292831718023e-05,
+      "loss": 2.5432,
+      "step": 3845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1078895798158717e-05,
+      "loss": 2.4927,
+      "step": 3846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107486310162166e-05,
+      "loss": 2.4849,
+      "step": 3847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107083022823257e-05,
+      "loss": 2.4984,
+      "step": 3848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1066797178654997e-05,
+      "loss": 2.5638,
+      "step": 3849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106276395355252e-05,
+      "loss": 2.4682,
+      "step": 3850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1058730553588735e-05,
+      "loss": 2.4821,
+      "step": 3851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105469697942728e-05,
+      "loss": 2.4817,
+      "step": 3852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105066323173182e-05,
+      "loss": 2.5941,
+      "step": 3853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1046629311166042e-05,
+      "loss": 2.5488,
+      "step": 3854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042595218393662e-05,
+      "loss": 2.4212,
+      "step": 3855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1038560954078434e-05,
+      "loss": 2.4988,
+      "step": 3856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1034526518884129e-05,
+      "loss": 2.504,
+      "step": 3857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1030491913474553e-05,
+      "loss": 2.4678,
+      "step": 3858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1026457138513535e-05,
+      "loss": 2.4428,
+      "step": 3859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1022422194664939e-05,
+      "loss": 2.435,
+      "step": 3860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1018387082592648e-05,
+      "loss": 2.4181,
+      "step": 3861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1014351802960576e-05,
+      "loss": 2.5019,
+      "step": 3862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1010316356432673e-05,
+      "loss": 2.5307,
+      "step": 3863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1006280743672903e-05,
+      "loss": 2.3886,
+      "step": 3864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100224496534526e-05,
+      "loss": 2.529,
+      "step": 3865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099820902211378e-05,
+      "loss": 2.5355,
+      "step": 3866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0994172914642504e-05,
+      "loss": 2.4903,
+      "step": 3867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0990136643595518e-05,
+      "loss": 2.5202,
+      "step": 3868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0986100209636922e-05,
+      "loss": 2.4574,
+      "step": 3869
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0982063613430854e-05,
+      "loss": 2.4527,
+      "step": 3870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097802685564147e-05,
+      "loss": 2.4618,
+      "step": 3871
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0973989936932952e-05,
+      "loss": 2.5796,
+      "step": 3872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0969952857969519e-05,
+      "loss": 2.5128,
+      "step": 3873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0965915619415408e-05,
+      "loss": 2.4934,
+      "step": 3874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0961878221934881e-05,
+      "loss": 2.5163,
+      "step": 3875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0957840666192233e-05,
+      "loss": 2.5023,
+      "step": 3876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953802952851778e-05,
+      "loss": 2.4804,
+      "step": 3877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0949765082577861e-05,
+      "loss": 2.4257,
+      "step": 3878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094572705603485e-05,
+      "loss": 2.542,
+      "step": 3879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0941688873887137e-05,
+      "loss": 2.488,
+      "step": 3880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093765053679915e-05,
+      "loss": 2.5144,
+      "step": 3881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0933612045435323e-05,
+      "loss": 2.5266,
+      "step": 3882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0929573400460136e-05,
+      "loss": 2.5782,
+      "step": 3883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925534602538083e-05,
+      "loss": 2.5036,
+      "step": 3884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0921495652333685e-05,
+      "loss": 2.6199,
+      "step": 3885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091745655051149e-05,
+      "loss": 2.5607,
+      "step": 3886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0913417297736068e-05,
+      "loss": 2.3887,
+      "step": 3887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0909377894672013e-05,
+      "loss": 2.5081,
+      "step": 3888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0905338341983951e-05,
+      "loss": 2.507,
+      "step": 3889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0901298640336526e-05,
+      "loss": 2.5224,
+      "step": 3890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0897258790394405e-05,
+      "loss": 2.5419,
+      "step": 3891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0893218792822285e-05,
+      "loss": 2.5133,
+      "step": 3892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0889178648284886e-05,
+      "loss": 2.4838,
+      "step": 3893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0885138357446949e-05,
+      "loss": 2.5722,
+      "step": 3894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0881097920973243e-05,
+      "loss": 2.4841,
+      "step": 3895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0877057339528556e-05,
+      "loss": 2.5759,
+      "step": 3896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873016613777707e-05,
+      "loss": 2.4739,
+      "step": 3897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0868975744385533e-05,
+      "loss": 2.6323,
+      "step": 3898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0864934732016893e-05,
+      "loss": 2.5027,
+      "step": 3899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0860893577336678e-05,
+      "loss": 2.5228,
+      "step": 3900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0856852281009794e-05,
+      "loss": 2.5535,
+      "step": 3901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852810843701173e-05,
+      "loss": 2.5734,
+      "step": 3902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0848769266075773e-05,
+      "loss": 2.5165,
+      "step": 3903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0844727548798578e-05,
+      "loss": 2.5403,
+      "step": 3904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0840685692534576e-05,
+      "loss": 2.4938,
+      "step": 3905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0836643697948802e-05,
+      "loss": 2.4729,
+      "step": 3906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0832601565706302e-05,
+      "loss": 2.5743,
+      "step": 3907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828559296472148e-05,
+      "loss": 2.4086,
+      "step": 3908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0824516890911426e-05,
+      "loss": 2.4574,
+      "step": 3909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820474349689254e-05,
+      "loss": 2.4935,
+      "step": 3910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0816431673470774e-05,
+      "loss": 2.4498,
+      "step": 3911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0812388862921138e-05,
+      "loss": 2.4965,
+      "step": 3912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0808345918705532e-05,
+      "loss": 2.5411,
+      "step": 3913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0804302841489159e-05,
+      "loss": 2.5387,
+      "step": 3914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0800259631937247e-05,
+      "loss": 2.484,
+      "step": 3915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0796216290715035e-05,
+      "loss": 2.571,
+      "step": 3916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0792172818487805e-05,
+      "loss": 2.5277,
+      "step": 3917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0788129215920837e-05,
+      "loss": 2.4207,
+      "step": 3918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0784085483679448e-05,
+      "loss": 2.5116,
+      "step": 3919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780041622428967e-05,
+      "loss": 2.5026,
+      "step": 3920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0775997632834756e-05,
+      "loss": 2.4186,
+      "step": 3921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0771953515562183e-05,
+      "loss": 2.5251,
+      "step": 3922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0767909271276649e-05,
+      "loss": 2.4639,
+      "step": 3923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076386490064357e-05,
+      "loss": 2.5102,
+      "step": 3924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0759820404328387e-05,
+      "loss": 2.5211,
+      "step": 3925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0755775782996553e-05,
+      "loss": 2.5846,
+      "step": 3926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751731037313555e-05,
+      "loss": 2.471,
+      "step": 3927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0747686167944889e-05,
+      "loss": 2.4841,
+      "step": 3928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0743641175556074e-05,
+      "loss": 2.5546,
+      "step": 3929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0739596060812653e-05,
+      "loss": 2.4479,
+      "step": 3930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0735550824380187e-05,
+      "loss": 2.4796,
+      "step": 3931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0731505466924258e-05,
+      "loss": 2.5092,
+      "step": 3932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0727459989110462e-05,
+      "loss": 2.4387,
+      "step": 3933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723414391604426e-05,
+      "loss": 2.4892,
+      "step": 3934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0719368675071785e-05,
+      "loss": 2.5454,
+      "step": 3935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07153228401782e-05,
+      "loss": 2.5614,
+      "step": 3936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071127688758935e-05,
+      "loss": 2.503,
+      "step": 3937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707230817970939e-05,
+      "loss": 2.479,
+      "step": 3938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0703184631988677e-05,
+      "loss": 2.442,
+      "step": 3939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0699138330308305e-05,
+      "loss": 2.4869,
+      "step": 3940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0695091913595579e-05,
+      "loss": 2.5292,
+      "step": 3941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691045382516277e-05,
+      "loss": 2.4566,
+      "step": 3942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0686998737736185e-05,
+      "loss": 2.4332,
+      "step": 3943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.068295197992112e-05,
+      "loss": 2.5478,
+      "step": 3944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678905109736914e-05,
+      "loss": 2.5005,
+      "step": 3945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0674858127849422e-05,
+      "loss": 2.4944,
+      "step": 3946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0670811034924498e-05,
+      "loss": 2.4417,
+      "step": 3947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066676383162804e-05,
+      "loss": 2.4172,
+      "step": 3948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0662716518625948e-05,
+      "loss": 2.5117,
+      "step": 3949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0658669096584142e-05,
+      "loss": 2.5001,
+      "step": 3950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.065462156616857e-05,
+      "loss": 2.5548,
+      "step": 3951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0650573928045185e-05,
+      "loss": 2.4199,
+      "step": 3952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0646526182879963e-05,
+      "loss": 2.4564,
+      "step": 3953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.06424783313389e-05,
+      "loss": 2.4749,
+      "step": 3954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0638430374088005e-05,
+      "loss": 2.5198,
+      "step": 3955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0634382311793306e-05,
+      "loss": 2.642,
+      "step": 3956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063033414512085e-05,
+      "loss": 2.5226,
+      "step": 3957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0626285874736699e-05,
+      "loss": 2.4985,
+      "step": 3958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0622237501306939e-05,
+      "loss": 2.428,
+      "step": 3959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0618189025497654e-05,
+      "loss": 2.533,
+      "step": 3960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0614140447974972e-05,
+      "loss": 2.4828,
+      "step": 3961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0610091769405013e-05,
+      "loss": 2.5334,
+      "step": 3962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0606042990453933e-05,
+      "loss": 2.4765,
+      "step": 3963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0601994111787886e-05,
+      "loss": 2.5071,
+      "step": 3964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059794513407306e-05,
+      "loss": 2.4869,
+      "step": 3965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0593896057975647e-05,
+      "loss": 2.4294,
+      "step": 3966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0589846884161863e-05,
+      "loss": 2.5433,
+      "step": 3967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0585797613297937e-05,
+      "loss": 2.4055,
+      "step": 3968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0581748246050107e-05,
+      "loss": 2.4966,
+      "step": 3969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0577698783084638e-05,
+      "loss": 2.4207,
+      "step": 3970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0573649225067808e-05,
+      "loss": 2.4934,
+      "step": 3971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056959957266591e-05,
+      "loss": 2.5348,
+      "step": 3972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0565549826545243e-05,
+      "loss": 2.4415,
+      "step": 3973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0561499987372137e-05,
+      "loss": 2.4847,
+      "step": 3974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0557450055812927e-05,
+      "loss": 2.4949,
+      "step": 3975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0553400032533974e-05,
+      "loss": 2.5235,
+      "step": 3976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549349918201633e-05,
+      "loss": 2.5024,
+      "step": 3977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05452997134823e-05,
+      "loss": 2.6363,
+      "step": 3978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0541249419042365e-05,
+      "loss": 2.4991,
+      "step": 3979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0537199035548245e-05,
+      "loss": 2.4647,
+      "step": 3980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0533148563666364e-05,
+      "loss": 2.551,
+      "step": 3981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0529098004063171e-05,
+      "loss": 2.5195,
+      "step": 3982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0525047357405115e-05,
+      "loss": 2.4152,
+      "step": 3983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0520996624358672e-05,
+      "loss": 2.4919,
+      "step": 3984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0516945805590326e-05,
+      "loss": 2.5096,
+      "step": 3985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512894901766579e-05,
+      "loss": 2.4938,
+      "step": 3986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0508843913553939e-05,
+      "loss": 2.5482,
+      "step": 3987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0504792841618934e-05,
+      "loss": 2.4956,
+      "step": 3988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050074168662811e-05,
+      "loss": 2.5547,
+      "step": 3989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0496690449248015e-05,
+      "loss": 2.5808,
+      "step": 3990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0492639130145222e-05,
+      "loss": 2.5184,
+      "step": 3991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048858772998631e-05,
+      "loss": 2.5678,
+      "step": 3992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0484536249437877e-05,
+      "loss": 2.467,
+      "step": 3993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0480484689166526e-05,
+      "loss": 2.5379,
+      "step": 3994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476433049838885e-05,
+      "loss": 2.6155,
+      "step": 3995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0472381332121584e-05,
+      "loss": 2.38,
+      "step": 3996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046832953668127e-05,
+      "loss": 2.429,
+      "step": 3997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0464277664184603e-05,
+      "loss": 2.5695,
+      "step": 3998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046022571529826e-05,
+      "loss": 2.5961,
+      "step": 3999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0456173690688922e-05,
+      "loss": 2.4502,
+      "step": 4000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0452121591023287e-05,
+      "loss": 2.5821,
+      "step": 4001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0448069416968065e-05,
+      "loss": 2.5091,
+      "step": 4002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444017169189985e-05,
+      "loss": 2.5359,
+      "step": 4003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0439964848355769e-05,
+      "loss": 2.501,
+      "step": 4004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043591245513217e-05,
+      "loss": 2.485,
+      "step": 4005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043185999018595e-05,
+      "loss": 2.5647,
+      "step": 4006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427807454183875e-05,
+      "loss": 2.4285,
+      "step": 4007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423754847792727e-05,
+      "loss": 2.5064,
+      "step": 4008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04197021716793e-05,
+      "loss": 2.4927,
+      "step": 4009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04156494265104e-05,
+      "loss": 2.4601,
+      "step": 4010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411596612952837e-05,
+      "loss": 2.538,
+      "step": 4011
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407543731673449e-05,
+      "loss": 2.4561,
+      "step": 4012
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0403490783339067e-05,
+      "loss": 2.5289,
+      "step": 4013
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0399437768616543e-05,
+      "loss": 2.5646,
+      "step": 4014
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0395384688172738e-05,
+      "loss": 2.4736,
+      "step": 4015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0391331542674526e-05,
+      "loss": 2.4934,
+      "step": 4016
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0387278332788782e-05,
+      "loss": 2.5408,
+      "step": 4017
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0383225059182405e-05,
+      "loss": 2.4709,
+      "step": 4018
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0379171722522295e-05,
+      "loss": 2.4804,
+      "step": 4019
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0375118323475368e-05,
+      "loss": 2.533,
+      "step": 4020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0371064862708547e-05,
+      "loss": 2.4794,
+      "step": 4021
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0367011340888765e-05,
+      "loss": 2.4792,
+      "step": 4022
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0362957758682965e-05,
+      "loss": 2.4892,
+      "step": 4023
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0358904116758106e-05,
+      "loss": 2.4596,
+      "step": 4024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0354850415781146e-05,
+      "loss": 2.4863,
+      "step": 4025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350796656419062e-05,
+      "loss": 2.5362,
+      "step": 4026
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0346742839338839e-05,
+      "loss": 2.5007,
+      "step": 4027
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0342688965207461e-05,
+      "loss": 2.4644,
+      "step": 4028
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0338635034691939e-05,
+      "loss": 2.4841,
+      "step": 4029
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0334581048459282e-05,
+      "loss": 2.527,
+      "step": 4030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0330527007176508e-05,
+      "loss": 2.5798,
+      "step": 4031
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.032647291151065e-05,
+      "loss": 2.4555,
+      "step": 4032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0322418762128745e-05,
+      "loss": 2.5154,
+      "step": 4033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0318364559697837e-05,
+      "loss": 2.4248,
+      "step": 4034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0314310304884987e-05,
+      "loss": 2.4947,
+      "step": 4035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0310255998357257e-05,
+      "loss": 2.4761,
+      "step": 4036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0306201640781723e-05,
+      "loss": 2.5372,
+      "step": 4037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302147232825463e-05,
+      "loss": 2.4559,
+      "step": 4038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0298092775155568e-05,
+      "loss": 2.6093,
+      "step": 4039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294038268439138e-05,
+      "loss": 2.4504,
+      "step": 4040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0289983713343277e-05,
+      "loss": 2.5402,
+      "step": 4041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0285929110535096e-05,
+      "loss": 2.5246,
+      "step": 4042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0281874460681724e-05,
+      "loss": 2.4579,
+      "step": 4043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0277819764450285e-05,
+      "loss": 2.5339,
+      "step": 4044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027376502250792e-05,
+      "loss": 2.5044,
+      "step": 4045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0269710235521775e-05,
+      "loss": 2.5448,
+      "step": 4046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0265655404158997e-05,
+      "loss": 2.56,
+      "step": 4047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0261600529086745e-05,
+      "loss": 2.3821,
+      "step": 4048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0257545610972191e-05,
+      "loss": 2.4606,
+      "step": 4049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025349065048251e-05,
+      "loss": 2.4089,
+      "step": 4050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0249435648284875e-05,
+      "loss": 2.4632,
+      "step": 4051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0245380605046478e-05,
+      "loss": 2.496,
+      "step": 4052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0241325521434515e-05,
+      "loss": 2.499,
+      "step": 4053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0237270398116186e-05,
+      "loss": 2.5243,
+      "step": 4054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0233215235758693e-05,
+      "loss": 2.4286,
+      "step": 4055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229160035029262e-05,
+      "loss": 2.5829,
+      "step": 4056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0225104796595102e-05,
+      "loss": 2.5383,
+      "step": 4057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0221049521123444e-05,
+      "loss": 2.4585,
+      "step": 4058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0216994209281522e-05,
+      "loss": 2.5591,
+      "step": 4059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0212938861736573e-05,
+      "loss": 2.4208,
+      "step": 4060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0208883479155838e-05,
+      "loss": 2.5467,
+      "step": 4061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0204828062206573e-05,
+      "loss": 2.5701,
+      "step": 4062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0200772611556029e-05,
+      "loss": 2.4331,
+      "step": 4063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0196717127871473e-05,
+      "loss": 2.4927,
+      "step": 4064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019266161182017e-05,
+      "loss": 2.5883,
+      "step": 4065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0188606064069387e-05,
+      "loss": 2.497,
+      "step": 4066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0184550485286408e-05,
+      "loss": 2.4455,
+      "step": 4067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0180494876138511e-05,
+      "loss": 2.4932,
+      "step": 4068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0176439237292985e-05,
+      "loss": 2.5168,
+      "step": 4069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0172383569417124e-05,
+      "loss": 2.532,
+      "step": 4070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0168327873178225e-05,
+      "loss": 2.5174,
+      "step": 4071
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0164272149243585e-05,
+      "loss": 2.5725,
+      "step": 4072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0160216398280518e-05,
+      "loss": 2.4501,
+      "step": 4073
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015616062095633e-05,
+      "loss": 2.4614,
+      "step": 4074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0152104817938336e-05,
+      "loss": 2.4848,
+      "step": 4075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0148048989893854e-05,
+      "loss": 2.5504,
+      "step": 4076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0143993137490216e-05,
+      "loss": 2.5236,
+      "step": 4077
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0139937261394739e-05,
+      "loss": 2.5495,
+      "step": 4078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013588136227476e-05,
+      "loss": 2.4931,
+      "step": 4079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0131825440797613e-05,
+      "loss": 2.4967,
+      "step": 4080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012776949763064e-05,
+      "loss": 2.4704,
+      "step": 4081
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123713533441177e-05,
+      "loss": 2.5462,
+      "step": 4082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0119657548896574e-05,
+      "loss": 2.5822,
+      "step": 4083
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0115601544664182e-05,
+      "loss": 2.5526,
+      "step": 4084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0111545521411347e-05,
+      "loss": 2.5165,
+      "step": 4085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0107489479805432e-05,
+      "loss": 2.5402,
+      "step": 4086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.010343342051379e-05,
+      "loss": 2.4881,
+      "step": 4087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099377344203789e-05,
+      "loss": 2.4797,
+      "step": 4088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0095321251542785e-05,
+      "loss": 2.4224,
+      "step": 4089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0091265143198152e-05,
+      "loss": 2.6217,
+      "step": 4090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087209019837256e-05,
+      "loss": 2.4864,
+      "step": 4091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.008315288212747e-05,
+      "loss": 2.5765,
+      "step": 4092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.007909673073617e-05,
+      "loss": 2.4684,
+      "step": 4093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0075040566330732e-05,
+      "loss": 2.5303,
+      "step": 4094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.007098438957853e-05,
+      "loss": 2.5543,
+      "step": 4095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0066928201146952e-05,
+      "loss": 2.5867,
+      "step": 4096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062872001703378e-05,
+      "loss": 2.5515,
+      "step": 4097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058815791915196e-05,
+      "loss": 2.5341,
+      "step": 4098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0054759572449784e-05,
+      "loss": 2.5002,
+      "step": 4099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050703343974535e-05,
+      "loss": 2.5175,
+      "step": 4100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046647107156839e-05,
+      "loss": 2.5081,
+      "step": 4101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042590862664089e-05,
+      "loss": 2.5251,
+      "step": 4102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0038534611163669e-05,
+      "loss": 2.4767,
+      "step": 4103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0034478353322979e-05,
+      "loss": 2.5614,
+      "step": 4104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030422089809416e-05,
+      "loss": 2.5243,
+      "step": 4105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026365821290363e-05,
+      "loss": 2.4984,
+      "step": 4106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0022309548433227e-05,
+      "loss": 2.533,
+      "step": 4107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0018253271905401e-05,
+      "loss": 2.3825,
+      "step": 4108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0014196992374283e-05,
+      "loss": 2.5092,
+      "step": 4109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001014071050727e-05,
+      "loss": 2.5126,
+      "step": 4110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0006084426971763e-05,
+      "loss": 2.5162,
+      "step": 4111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0002028142435155e-05,
+      "loss": 2.4905,
+      "step": 4112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997971857564847e-06,
+      "loss": 2.45,
+      "step": 4113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99391557302824e-06,
+      "loss": 2.5296,
+      "step": 4114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.989859289492732e-06,
+      "loss": 2.577,
+      "step": 4115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985803007625719e-06,
+      "loss": 2.5263,
+      "step": 4116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981746728094599e-06,
+      "loss": 2.5336,
+      "step": 4117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.977690451566775e-06,
+      "loss": 2.5225,
+      "step": 4118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97363417870964e-06,
+      "loss": 2.5415,
+      "step": 4119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96957791019059e-06,
+      "loss": 2.5128,
+      "step": 4120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.965521646677021e-06,
+      "loss": 2.5068,
+      "step": 4121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.961465388836333e-06,
+      "loss": 2.5284,
+      "step": 4122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.957409137335915e-06,
+      "loss": 2.4825,
+      "step": 4123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.953352892843165e-06,
+      "loss": 2.4476,
+      "step": 4124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94929665602547e-06,
+      "loss": 2.4917,
+      "step": 4125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94524042755022e-06,
+      "loss": 2.4716,
+      "step": 4126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94118420808481e-06,
+      "loss": 2.5037,
+      "step": 4127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937127998296625e-06,
+      "loss": 2.4689,
+      "step": 4128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93307179885305e-06,
+      "loss": 2.5009,
+      "step": 4129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.92901561042147e-06,
+      "loss": 2.5266,
+      "step": 4130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.924959433669272e-06,
+      "loss": 2.5981,
+      "step": 4131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920903269263834e-06,
+      "loss": 2.4208,
+      "step": 4132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.916847117872532e-06,
+      "loss": 2.604,
+      "step": 4133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912790980162744e-06,
+      "loss": 2.4869,
+      "step": 4134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.908734856801851e-06,
+      "loss": 2.5163,
+      "step": 4135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90467874845722e-06,
+      "loss": 2.4384,
+      "step": 4136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900622655796216e-06,
+      "loss": 2.5173,
+      "step": 4137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89656657948621e-06,
+      "loss": 2.4294,
+      "step": 4138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.892510520194571e-06,
+      "loss": 2.418,
+      "step": 4139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.888454478588655e-06,
+      "loss": 2.4934,
+      "step": 4140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.884398455335823e-06,
+      "loss": 2.4062,
+      "step": 4141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880342451103431e-06,
+      "loss": 2.5006,
+      "step": 4142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.876286466558826e-06,
+      "loss": 2.4993,
+      "step": 4143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.872230502369363e-06,
+      "loss": 2.5194,
+      "step": 4144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.868174559202389e-06,
+      "loss": 2.3639,
+      "step": 4145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864118637725241e-06,
+      "loss": 2.5326,
+      "step": 4146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.860062738605261e-06,
+      "loss": 2.4604,
+      "step": 4147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856006862509787e-06,
+      "loss": 2.4755,
+      "step": 4148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.851951010106147e-06,
+      "loss": 2.3819,
+      "step": 4149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847895182061667e-06,
+      "loss": 2.4806,
+      "step": 4150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.843839379043672e-06,
+      "loss": 2.4554,
+      "step": 4151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839783601719485e-06,
+      "loss": 2.5278,
+      "step": 4152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.835727850756416e-06,
+      "loss": 2.5108,
+      "step": 4153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.831672126821779e-06,
+      "loss": 2.5089,
+      "step": 4154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827616430582876e-06,
+      "loss": 2.5449,
+      "step": 4155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.823560762707017e-06,
+      "loss": 2.4505,
+      "step": 4156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81950512386149e-06,
+      "loss": 2.5143,
+      "step": 4157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.815449514713597e-06,
+      "loss": 2.5609,
+      "step": 4158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811393935930613e-06,
+      "loss": 2.5442,
+      "step": 4159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.807338388179835e-06,
+      "loss": 2.5548,
+      "step": 4160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.803282872128528e-06,
+      "loss": 2.5027,
+      "step": 4161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.799227388443973e-06,
+      "loss": 2.5383,
+      "step": 4162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79517193779343e-06,
+      "loss": 2.5067,
+      "step": 4163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.791116520844163e-06,
+      "loss": 2.4069,
+      "step": 4164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.787061138263432e-06,
+      "loss": 2.5266,
+      "step": 4165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.783005790718483e-06,
+      "loss": 2.5542,
+      "step": 4166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.778950478876559e-06,
+      "loss": 2.4913,
+      "step": 4167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.7748952034049e-06,
+      "loss": 2.4894,
+      "step": 4168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770839964970742e-06,
+      "loss": 2.4929,
+      "step": 4169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76678476424131e-06,
+      "loss": 2.4887,
+      "step": 4170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76272960188382e-06,
+      "loss": 2.4977,
+      "step": 4171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758674478565486e-06,
+      "loss": 2.5587,
+      "step": 4172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.754619394953523e-06,
+      "loss": 2.5359,
+      "step": 4173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750564351715128e-06,
+      "loss": 2.4863,
+      "step": 4174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746509349517495e-06,
+      "loss": 2.4674,
+      "step": 4175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.742454389027809e-06,
+      "loss": 2.459,
+      "step": 4176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738399470913256e-06,
+      "loss": 2.4956,
+      "step": 4177
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.734344595841008e-06,
+      "loss": 2.5127,
+      "step": 4178
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73028976447823e-06,
+      "loss": 2.5657,
+      "step": 4179
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.726234977492079e-06,
+      "loss": 2.4864,
+      "step": 4180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.722180235549714e-06,
+      "loss": 2.5023,
+      "step": 4181
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.718125539318277e-06,
+      "loss": 2.4823,
+      "step": 4182
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.714070889464907e-06,
+      "loss": 2.4461,
+      "step": 4183
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71001628665673e-06,
+      "loss": 2.5031,
+      "step": 4184
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.705961731560864e-06,
+      "loss": 2.5356,
+      "step": 4185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.701907224844435e-06,
+      "loss": 2.4995,
+      "step": 4186
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.69785276717454e-06,
+      "loss": 2.3913,
+      "step": 4187
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.693798359218282e-06,
+      "loss": 2.5252,
+      "step": 4188
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.689744001642744e-06,
+      "loss": 2.6266,
+      "step": 4189
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.685689695115016e-06,
+      "loss": 2.5197,
+      "step": 4190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.681635440302166e-06,
+      "loss": 2.515,
+      "step": 4191
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.677581237871261e-06,
+      "loss": 2.546,
+      "step": 4192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.673527088489351e-06,
+      "loss": 2.5148,
+      "step": 4193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669472992823494e-06,
+      "loss": 2.539,
+      "step": 4194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.665418951540721e-06,
+      "loss": 2.5498,
+      "step": 4195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.661364965308064e-06,
+      "loss": 2.4005,
+      "step": 4196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65731103479254e-06,
+      "loss": 2.4346,
+      "step": 4197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.653257160661164e-06,
+      "loss": 2.4796,
+      "step": 4198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64920334358094e-06,
+      "loss": 2.5037,
+      "step": 4199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.645149584218859e-06,
+      "loss": 2.3764,
+      "step": 4200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6410958832419e-06,
+      "loss": 2.5339,
+      "step": 4201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.637042241317035e-06,
+      "loss": 2.4885,
+      "step": 4202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.632988659111239e-06,
+      "loss": 2.5823,
+      "step": 4203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.628935137291457e-06,
+      "loss": 2.5224,
+      "step": 4204
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.624881676524635e-06,
+      "loss": 2.4052,
+      "step": 4205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.620828277477707e-06,
+      "loss": 2.414,
+      "step": 4206
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.616774940817599e-06,
+      "loss": 2.5,
+      "step": 4207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61272166721122e-06,
+      "loss": 2.5365,
+      "step": 4208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.60866845732548e-06,
+      "loss": 2.4856,
+      "step": 4209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604615311827262e-06,
+      "loss": 2.3864,
+      "step": 4210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.600562231383459e-06,
+      "loss": 2.4235,
+      "step": 4211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596509216660935e-06,
+      "loss": 2.4549,
+      "step": 4212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592456268326556e-06,
+      "loss": 2.3805,
+      "step": 4213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588403387047165e-06,
+      "loss": 2.4545,
+      "step": 4214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.584350573489604e-06,
+      "loss": 2.4312,
+      "step": 4215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.580297828320704e-06,
+      "loss": 2.4318,
+      "step": 4216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576245152207278e-06,
+      "loss": 2.5433,
+      "step": 4217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572192545816126e-06,
+      "loss": 2.4494,
+      "step": 4218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.568140009814051e-06,
+      "loss": 2.4504,
+      "step": 4219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.564087544867831e-06,
+      "loss": 2.4202,
+      "step": 4220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.560035151644235e-06,
+      "loss": 2.5341,
+      "step": 4221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555982830810022e-06,
+      "loss": 2.4494,
+      "step": 4222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551930583031935e-06,
+      "loss": 2.4592,
+      "step": 4223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.547878408976716e-06,
+      "loss": 2.5376,
+      "step": 4224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54382630931108e-06,
+      "loss": 2.4722,
+      "step": 4225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.539774284701743e-06,
+      "loss": 2.4219,
+      "step": 4226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.535722335815397e-06,
+      "loss": 2.5602,
+      "step": 4227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531670463318733e-06,
+      "loss": 2.4907,
+      "step": 4228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52761866787842e-06,
+      "loss": 2.4867,
+      "step": 4229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52356695016112e-06,
+      "loss": 2.4818,
+      "step": 4230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.519515310833475e-06,
+      "loss": 2.5116,
+      "step": 4231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515463750562125e-06,
+      "loss": 2.4991,
+      "step": 4232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511412270013693e-06,
+      "loss": 2.5623,
+      "step": 4233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507360869854783e-06,
+      "loss": 2.5382,
+      "step": 4234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.503309550751987e-06,
+      "loss": 2.5109,
+      "step": 4235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499258313371893e-06,
+      "loss": 2.535,
+      "step": 4236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49520715838107e-06,
+      "loss": 2.4737,
+      "step": 4237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.491156086446065e-06,
+      "loss": 2.4392,
+      "step": 4238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487105098233421e-06,
+      "loss": 2.4941,
+      "step": 4239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.483054194409674e-06,
+      "loss": 2.5726,
+      "step": 4240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47900337564133e-06,
+      "loss": 2.5313,
+      "step": 4241
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.474952642594886e-06,
+      "loss": 2.575,
+      "step": 4242
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.470901995936834e-06,
+      "loss": 2.498,
+      "step": 4243
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.466851436333636e-06,
+      "loss": 2.4171,
+      "step": 4244
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46280096445176e-06,
+      "loss": 2.5278,
+      "step": 4245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458750580957638e-06,
+      "loss": 2.3974,
+      "step": 4246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454700286517706e-06,
+      "loss": 2.584,
+      "step": 4247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.450650081798369e-06,
+      "loss": 2.5166,
+      "step": 4248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.446599967466029e-06,
+      "loss": 2.5434,
+      "step": 4249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442549944187075e-06,
+      "loss": 2.4669,
+      "step": 4250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.438500012627868e-06,
+      "loss": 2.4713,
+      "step": 4251
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43445017345476e-06,
+      "loss": 2.4776,
+      "step": 4252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430400427334094e-06,
+      "loss": 2.4702,
+      "step": 4253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426350774932195e-06,
+      "loss": 2.6179,
+      "step": 4254
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.422301216915364e-06,
+      "loss": 2.4988,
+      "step": 4255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.418251753949895e-06,
+      "loss": 2.5233,
+      "step": 4256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.414202386702068e-06,
+      "loss": 2.5359,
+      "step": 4257
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41015311583814e-06,
+      "loss": 2.4658,
+      "step": 4258
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.406103942024354e-06,
+      "loss": 2.4664,
+      "step": 4259
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402054865926944e-06,
+      "loss": 2.4625,
+      "step": 4260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398005888212116e-06,
+      "loss": 2.4895,
+      "step": 4261
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.393957009546069e-06,
+      "loss": 2.5468,
+      "step": 4262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389908230594989e-06,
+      "loss": 2.4425,
+      "step": 4263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.385859552025033e-06,
+      "loss": 2.4487,
+      "step": 4264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.381810974502347e-06,
+      "loss": 2.4967,
+      "step": 4265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377762498693065e-06,
+      "loss": 2.4792,
+      "step": 4266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373714125263303e-06,
+      "loss": 2.4421,
+      "step": 4267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.369665854879154e-06,
+      "loss": 2.3945,
+      "step": 4268
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.365617688206696e-06,
+      "loss": 2.5268,
+      "step": 4269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.361569625911998e-06,
+      "loss": 2.4485,
+      "step": 4270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357521668661104e-06,
+      "loss": 2.5595,
+      "step": 4271
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.353473817120039e-06,
+      "loss": 2.6193,
+      "step": 4272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.349426071954815e-06,
+      "loss": 2.4605,
+      "step": 4273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345378433831431e-06,
+      "loss": 2.5394,
+      "step": 4274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34133090341586e-06,
+      "loss": 2.4834,
+      "step": 4275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.337283481374057e-06,
+      "loss": 2.4331,
+      "step": 4276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.333236168371962e-06,
+      "loss": 2.488,
+      "step": 4277
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.329188965075504e-06,
+      "loss": 2.4276,
+      "step": 4278
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.325141872150583e-06,
+      "loss": 2.4911,
+      "step": 4279
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321094890263088e-06,
+      "loss": 2.4856,
+      "step": 4280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317048020078884e-06,
+      "loss": 2.4974,
+      "step": 4281
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313001262263818e-06,
+      "loss": 2.59,
+      "step": 4282
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308954617483728e-06,
+      "loss": 2.4981,
+      "step": 4283
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.304908086404425e-06,
+      "loss": 2.5138,
+      "step": 4284
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.3008616696917e-06,
+      "loss": 2.4748,
+      "step": 4285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.296815368011326e-06,
+      "loss": 2.5856,
+      "step": 4286
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.292769182029065e-06,
+      "loss": 2.5688,
+      "step": 4287
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.288723112410653e-06,
+      "loss": 2.3826,
+      "step": 4288
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.284677159821803e-06,
+      "loss": 2.4414,
+      "step": 4289
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.280631324928217e-06,
+      "loss": 2.4961,
+      "step": 4290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276585608395578e-06,
+      "loss": 2.4562,
+      "step": 4291
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.272540010889541e-06,
+      "loss": 2.54,
+      "step": 4292
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268494533075747e-06,
+      "loss": 2.465,
+      "step": 4293
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.264449175619813e-06,
+      "loss": 2.4787,
+      "step": 4294
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.260403939187349e-06,
+      "loss": 2.389,
+      "step": 4295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.256358824443929e-06,
+      "loss": 2.3799,
+      "step": 4296
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252313832055116e-06,
+      "loss": 2.4574,
+      "step": 4297
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.248268962686451e-06,
+      "loss": 2.5432,
+      "step": 4298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.244224217003449e-06,
+      "loss": 2.4798,
+      "step": 4299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240179595671615e-06,
+      "loss": 2.4936,
+      "step": 4300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.236135099356433e-06,
+      "loss": 2.5946,
+      "step": 4301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.232090728723353e-06,
+      "loss": 2.5705,
+      "step": 4302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.228046484437819e-06,
+      "loss": 2.5388,
+      "step": 4303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.224002367165246e-06,
+      "loss": 2.4825,
+      "step": 4304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219958377571035e-06,
+      "loss": 2.4987,
+      "step": 4305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.215914516320555e-06,
+      "loss": 2.4689,
+      "step": 4306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.211870784079163e-06,
+      "loss": 2.6097,
+      "step": 4307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.207827181512197e-06,
+      "loss": 2.4765,
+      "step": 4308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203783709284966e-06,
+      "loss": 2.5603,
+      "step": 4309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.199740368062758e-06,
+      "loss": 2.5389,
+      "step": 4310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195697158510843e-06,
+      "loss": 2.5355,
+      "step": 4311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.191654081294473e-06,
+      "loss": 2.5296,
+      "step": 4312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187611137078866e-06,
+      "loss": 2.4376,
+      "step": 4313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.183568326529233e-06,
+      "loss": 2.4629,
+      "step": 4314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179525650310747e-06,
+      "loss": 2.5706,
+      "step": 4315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.175483109088578e-06,
+      "loss": 2.5403,
+      "step": 4316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171440703527856e-06,
+      "loss": 2.4933,
+      "step": 4317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1673984342937e-06,
+      "loss": 2.4722,
+      "step": 4318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1633563020512e-06,
+      "loss": 2.3797,
+      "step": 4319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.159314307465425e-06,
+      "loss": 2.5172,
+      "step": 4320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.155272451201427e-06,
+      "loss": 2.5703,
+      "step": 4321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.151230733924228e-06,
+      "loss": 2.5008,
+      "step": 4322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147189156298828e-06,
+      "loss": 2.4931,
+      "step": 4323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.143147718990208e-06,
+      "loss": 2.5143,
+      "step": 4324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139106422663325e-06,
+      "loss": 2.5375,
+      "step": 4325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13506526798311e-06,
+      "loss": 2.4867,
+      "step": 4326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.131024255614472e-06,
+      "loss": 2.4123,
+      "step": 4327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126983386222295e-06,
+      "loss": 2.4018,
+      "step": 4328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.122942660471447e-06,
+      "loss": 2.415,
+      "step": 4329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11890207902676e-06,
+      "loss": 2.4152,
+      "step": 4330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.114861642553054e-06,
+      "loss": 2.4333,
+      "step": 4331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.110821351715115e-06,
+      "loss": 2.5167,
+      "step": 4332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.106781207177717e-06,
+      "loss": 2.5165,
+      "step": 4333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.102741209605598e-06,
+      "loss": 2.5126,
+      "step": 4334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09870135966348e-06,
+      "loss": 2.48,
+      "step": 4335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09466165801605e-06,
+      "loss": 2.5169,
+      "step": 4336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.090622105327987e-06,
+      "loss": 2.4591,
+      "step": 4337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.086582702263935e-06,
+      "loss": 2.5602,
+      "step": 4338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.082543449488514e-06,
+      "loss": 2.4622,
+      "step": 4339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078504347666317e-06,
+      "loss": 2.5372,
+      "step": 4340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074465397461917e-06,
+      "loss": 2.4235,
+      "step": 4341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.070426599539866e-06,
+      "loss": 2.5685,
+      "step": 4342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.066387954564682e-06,
+      "loss": 2.4691,
+      "step": 4343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.062349463200857e-06,
+      "loss": 2.498,
+      "step": 4344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.058311126112865e-06,
+      "loss": 2.4737,
+      "step": 4345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.054272943965154e-06,
+      "loss": 2.4599,
+      "step": 4346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050234917422142e-06,
+      "loss": 2.5205,
+      "step": 4347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046197047148226e-06,
+      "loss": 2.4642,
+      "step": 4348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.042159333807769e-06,
+      "loss": 2.4785,
+      "step": 4349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.038121778065122e-06,
+      "loss": 2.5494,
+      "step": 4350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.034084380584595e-06,
+      "loss": 2.4858,
+      "step": 4351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.030047142030485e-06,
+      "loss": 2.4703,
+      "step": 4352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02601006306705e-06,
+      "loss": 2.553,
+      "step": 4353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.021973144358533e-06,
+      "loss": 2.458,
+      "step": 4354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01793638656915e-06,
+      "loss": 2.4468,
+      "step": 4355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.013899790363081e-06,
+      "loss": 2.5838,
+      "step": 4356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.009863356404485e-06,
+      "loss": 2.4762,
+      "step": 4357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.005827085357495e-06,
+      "loss": 2.4547,
+      "step": 4358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.001790977886222e-06,
+      "loss": 2.5402,
+      "step": 4359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.997755034654741e-06,
+      "loss": 2.5052,
+      "step": 4360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.993719256327103e-06,
+      "loss": 2.5048,
+      "step": 4361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.989683643567329e-06,
+      "loss": 2.4825,
+      "step": 4362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.985648197039425e-06,
+      "loss": 2.4605,
+      "step": 4363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981612917407355e-06,
+      "loss": 2.4507,
+      "step": 4364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.977577805335068e-06,
+      "loss": 2.4761,
+      "step": 4365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973542861486466e-06,
+      "loss": 2.4633,
+      "step": 4366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.969508086525452e-06,
+      "loss": 2.5201,
+      "step": 4367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.965473481115873e-06,
+      "loss": 2.5025,
+      "step": 4368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961439045921571e-06,
+      "loss": 2.5614,
+      "step": 4369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95740478160634e-06,
+      "loss": 2.5239,
+      "step": 4370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.953370688833962e-06,
+      "loss": 2.4193,
+      "step": 4371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949336768268185e-06,
+      "loss": 2.3889,
+      "step": 4372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.945303020572725e-06,
+      "loss": 2.4473,
+      "step": 4373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941269446411268e-06,
+      "loss": 2.4394,
+      "step": 4374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.937236046447484e-06,
+      "loss": 2.5051,
+      "step": 4375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.933202821345004e-06,
+      "loss": 2.4413,
+      "step": 4376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.929169771767434e-06,
+      "loss": 2.4288,
+      "step": 4377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.925136898378343e-06,
+      "loss": 2.5602,
+      "step": 4378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.921104201841281e-06,
+      "loss": 2.4913,
+      "step": 4379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917071682819773e-06,
+      "loss": 2.3667,
+      "step": 4380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913039341977297e-06,
+      "loss": 2.4903,
+      "step": 4381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90900717997732e-06,
+      "loss": 2.3744,
+      "step": 4382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90497519748326e-06,
+      "loss": 2.4645,
+      "step": 4383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.900943395158533e-06,
+      "loss": 2.4938,
+      "step": 4384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.896911773666499e-06,
+      "loss": 2.4878,
+      "step": 4385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8928803336705e-06,
+      "loss": 2.4545,
+      "step": 4386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.888849075833848e-06,
+      "loss": 2.5306,
+      "step": 4387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.884818000819824e-06,
+      "loss": 2.5532,
+      "step": 4388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88078710929168e-06,
+      "loss": 2.5252,
+      "step": 4389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876756401912636e-06,
+      "loss": 2.4138,
+      "step": 4390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872725879345878e-06,
+      "loss": 2.5622,
+      "step": 4391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86869554225457e-06,
+      "loss": 2.5674,
+      "step": 4392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.864665391301843e-06,
+      "loss": 2.4479,
+      "step": 4393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.860635427150792e-06,
+      "loss": 2.3627,
+      "step": 4394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856605650464487e-06,
+      "loss": 2.4852,
+      "step": 4395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852576061905964e-06,
+      "loss": 2.4736,
+      "step": 4396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848546662138233e-06,
+      "loss": 2.4948,
+      "step": 4397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.844517451824264e-06,
+      "loss": 2.529,
+      "step": 4398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.840488431627002e-06,
+      "loss": 2.4601,
+      "step": 4399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83645960220936e-06,
+      "loss": 2.4925,
+      "step": 4400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832430964234224e-06,
+      "loss": 2.5148,
+      "step": 4401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.828402518364437e-06,
+      "loss": 2.5248,
+      "step": 4402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.824374265262821e-06,
+      "loss": 2.5524,
+      "step": 4403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820346205592159e-06,
+      "loss": 2.4675,
+      "step": 4404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81631834001521e-06,
+      "loss": 2.5239,
+      "step": 4405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.812290669194697e-06,
+      "loss": 2.5511,
+      "step": 4406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.808263193793309e-06,
+      "loss": 2.5427,
+      "step": 4407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8042359144737e-06,
+      "loss": 2.4905,
+      "step": 4408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.800208831898502e-06,
+      "loss": 2.3955,
+      "step": 4409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79618194673031e-06,
+      "loss": 2.4176,
+      "step": 4410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.792155259631679e-06,
+      "loss": 2.6322,
+      "step": 4411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.788128771265142e-06,
+      "loss": 2.4626,
+      "step": 4412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784102482293198e-06,
+      "loss": 2.4143,
+      "step": 4413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.780076393378306e-06,
+      "loss": 2.5423,
+      "step": 4414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.776050505182896e-06,
+      "loss": 2.5747,
+      "step": 4415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.772024818369368e-06,
+      "loss": 2.5407,
+      "step": 4416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.767999333600083e-06,
+      "loss": 2.4703,
+      "step": 4417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763974051537377e-06,
+      "loss": 2.4788,
+      "step": 4418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.759948972843544e-06,
+      "loss": 2.5458,
+      "step": 4419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755924098180849e-06,
+      "loss": 2.4718,
+      "step": 4420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.751899428211519e-06,
+      "loss": 2.4608,
+      "step": 4421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747874963597756e-06,
+      "loss": 2.5504,
+      "step": 4422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.743850705001725e-06,
+      "loss": 2.5026,
+      "step": 4423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739826653085552e-06,
+      "loss": 2.4015,
+      "step": 4424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.735802808511326e-06,
+      "loss": 2.5072,
+      "step": 4425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.731779171941118e-06,
+      "loss": 2.5437,
+      "step": 4426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.727755744036953e-06,
+      "loss": 2.3899,
+      "step": 4427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.723732525460819e-06,
+      "loss": 2.3852,
+      "step": 4428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.719709516874675e-06,
+      "loss": 2.5323,
+      "step": 4429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715686718940451e-06,
+      "loss": 2.4358,
+      "step": 4430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.711664132320031e-06,
+      "loss": 2.4825,
+      "step": 4431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.707641757675267e-06,
+      "loss": 2.4343,
+      "step": 4432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70361959566798e-06,
+      "loss": 2.4388,
+      "step": 4433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.699597646959958e-06,
+      "loss": 2.526,
+      "step": 4434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.695575912212948e-06,
+      "loss": 2.5628,
+      "step": 4435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69155439208866e-06,
+      "loss": 2.5706,
+      "step": 4436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68753308724878e-06,
+      "loss": 2.4655,
+      "step": 4437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68351199835494e-06,
+      "loss": 2.5924,
+      "step": 4438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67949112606876e-06,
+      "loss": 2.4829,
+      "step": 4439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675470471051808e-06,
+      "loss": 2.5812,
+      "step": 4440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67145003396562e-06,
+      "loss": 2.582,
+      "step": 4441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66742981547169e-06,
+      "loss": 2.5073,
+      "step": 4442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663409816231492e-06,
+      "loss": 2.4649,
+      "step": 4443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659390036906452e-06,
+      "loss": 2.4314,
+      "step": 4444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65537047815796e-06,
+      "loss": 2.4824,
+      "step": 4445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65135114064737e-06,
+      "loss": 2.3767,
+      "step": 4446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64733202503601e-06,
+      "loss": 2.5076,
+      "step": 4447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.643313131985156e-06,
+      "loss": 2.4504,
+      "step": 4448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.639294462156056e-06,
+      "loss": 2.4531,
+      "step": 4449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.635276016209916e-06,
+      "loss": 2.4869,
+      "step": 4450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.631257794807919e-06,
+      "loss": 2.5059,
+      "step": 4451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.627239798611192e-06,
+      "loss": 2.3731,
+      "step": 4452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623222028280835e-06,
+      "loss": 2.5556,
+      "step": 4453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.619204484477914e-06,
+      "loss": 2.4752,
+      "step": 4454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.615187167863446e-06,
+      "loss": 2.562,
+      "step": 4455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.611170079098425e-06,
+      "loss": 2.5071,
+      "step": 4456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6071532188438e-06,
+      "loss": 2.459,
+      "step": 4457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.603136587760481e-06,
+      "loss": 2.515,
+      "step": 4458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59912018650934e-06,
+      "loss": 2.4511,
+      "step": 4459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595104015751217e-06,
+      "loss": 2.4325,
+      "step": 4460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59108807614691e-06,
+      "loss": 2.4644,
+      "step": 4461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58707236835718e-06,
+      "loss": 2.4937,
+      "step": 4462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583056893042742e-06,
+      "loss": 2.5046,
+      "step": 4463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.579041650864293e-06,
+      "loss": 2.4873,
+      "step": 4464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.575026642482474e-06,
+      "loss": 2.5299,
+      "step": 4465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.571011868557886e-06,
+      "loss": 2.3559,
+      "step": 4466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5669973297511e-06,
+      "loss": 2.4506,
+      "step": 4467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.562983026722654e-06,
+      "loss": 2.4699,
+      "step": 4468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55896896013303e-06,
+      "loss": 2.3701,
+      "step": 4469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.554955130642684e-06,
+      "loss": 2.5273,
+      "step": 4470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550941538912028e-06,
+      "loss": 2.5661,
+      "step": 4471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54692818560144e-06,
+      "loss": 2.5515,
+      "step": 4472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.542915071371249e-06,
+      "loss": 2.5245,
+      "step": 4473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.538902196881758e-06,
+      "loss": 2.5109,
+      "step": 4474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534889562793217e-06,
+      "loss": 2.49,
+      "step": 4475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53087716976584e-06,
+      "loss": 2.4799,
+      "step": 4476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526865018459813e-06,
+      "loss": 2.4691,
+      "step": 4477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522853109535269e-06,
+      "loss": 2.4034,
+      "step": 4478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.518841443652304e-06,
+      "loss": 2.516,
+      "step": 4479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.514830021470973e-06,
+      "loss": 2.52,
+      "step": 4480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.510818843651301e-06,
+      "loss": 2.4767,
+      "step": 4481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50680791085326e-06,
+      "loss": 2.482,
+      "step": 4482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502797223736786e-06,
+      "loss": 2.4692,
+      "step": 4483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.498786782961775e-06,
+      "loss": 2.5025,
+      "step": 4484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.494776589188088e-06,
+      "loss": 2.461,
+      "step": 4485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.490766643075534e-06,
+      "loss": 2.4183,
+      "step": 4486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486756945283891e-06,
+      "loss": 2.4937,
+      "step": 4487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.482747496472888e-06,
+      "loss": 2.4531,
+      "step": 4488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478738297302225e-06,
+      "loss": 2.4234,
+      "step": 4489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.474729348431548e-06,
+      "loss": 2.4863,
+      "step": 4490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.470720650520468e-06,
+      "loss": 2.4803,
+      "step": 4491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.466712204228552e-06,
+      "loss": 2.5979,
+      "step": 4492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.462704010215335e-06,
+      "loss": 2.5106,
+      "step": 4493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.458696069140294e-06,
+      "loss": 2.6316,
+      "step": 4494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.454688381662881e-06,
+      "loss": 2.52,
+      "step": 4495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.450680948442494e-06,
+      "loss": 2.48,
+      "step": 4496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.446673770138491e-06,
+      "loss": 2.4855,
+      "step": 4497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.442666847410202e-06,
+      "loss": 2.5168,
+      "step": 4498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438660180916895e-06,
+      "loss": 2.4586,
+      "step": 4499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.434653771317804e-06,
+      "loss": 2.5101,
+      "step": 4500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430647619272124e-06,
+      "loss": 2.5013,
+      "step": 4501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426641725439008e-06,
+      "loss": 2.5741,
+      "step": 4502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42263609047756e-06,
+      "loss": 2.5156,
+      "step": 4503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41863071504685e-06,
+      "loss": 2.565,
+      "step": 4504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.414625599805891e-06,
+      "loss": 2.4371,
+      "step": 4505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41062074541367e-06,
+      "loss": 2.5035,
+      "step": 4506
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.406616152529121e-06,
+      "loss": 2.4484,
+      "step": 4507
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40261182181114e-06,
+      "loss": 2.4262,
+      "step": 4508
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39860775391857e-06,
+      "loss": 2.5317,
+      "step": 4509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39460394951023e-06,
+      "loss": 2.5682,
+      "step": 4510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390600409244874e-06,
+      "loss": 2.4835,
+      "step": 4511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38659713378123e-06,
+      "loss": 2.5109,
+      "step": 4512
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.382594123777965e-06,
+      "loss": 2.4709,
+      "step": 4513
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.378591379893718e-06,
+      "loss": 2.5186,
+      "step": 4514
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374588902787082e-06,
+      "loss": 2.4792,
+      "step": 4515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.370586693116599e-06,
+      "loss": 2.5184,
+      "step": 4516
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.366584751540767e-06,
+      "loss": 2.5734,
+      "step": 4517
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.362583078718045e-06,
+      "loss": 2.6245,
+      "step": 4518
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35858167530685e-06,
+      "loss": 2.4706,
+      "step": 4519
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.354580541965549e-06,
+      "loss": 2.4287,
+      "step": 4520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.350579679352465e-06,
+      "loss": 2.5157,
+      "step": 4521
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.346579088125877e-06,
+      "loss": 2.534,
+      "step": 4522
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.342578768944025e-06,
+      "loss": 2.4635,
+      "step": 4523
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.338578722465092e-06,
+      "loss": 2.5073,
+      "step": 4524
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.334578949347232e-06,
+      "loss": 2.5387,
+      "step": 4525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.330579450248536e-06,
+      "loss": 2.557,
+      "step": 4526
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.326580225827067e-06,
+      "loss": 2.466,
+      "step": 4527
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.322581276740833e-06,
+      "loss": 2.4773,
+      "step": 4528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3185826036478e-06,
+      "loss": 2.3291,
+      "step": 4529
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.314584207205881e-06,
+      "loss": 2.5269,
+      "step": 4530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.310586088072958e-06,
+      "loss": 2.5343,
+      "step": 4531
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30658824690686e-06,
+      "loss": 2.4102,
+      "step": 4532
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.302590684365366e-06,
+      "loss": 2.4464,
+      "step": 4533
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.298593401106211e-06,
+      "loss": 2.5639,
+      "step": 4534
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294596397787086e-06,
+      "loss": 2.5418,
+      "step": 4535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.290599675065642e-06,
+      "loss": 2.3385,
+      "step": 4536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286603233599472e-06,
+      "loss": 2.4578,
+      "step": 4537
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.282607074046131e-06,
+      "loss": 2.5459,
+      "step": 4538
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.27861119706312e-06,
+      "loss": 2.4334,
+      "step": 4539
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274615603307907e-06,
+      "loss": 2.5477,
+      "step": 4540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270620293437898e-06,
+      "loss": 2.4909,
+      "step": 4541
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266625268110464e-06,
+      "loss": 2.5067,
+      "step": 4542
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262630527982919e-06,
+      "loss": 2.5159,
+      "step": 4543
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25863607371254e-06,
+      "loss": 2.4426,
+      "step": 4544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.254641905956551e-06,
+      "loss": 2.4494,
+      "step": 4545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250648025372132e-06,
+      "loss": 2.4706,
+      "step": 4546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246654432616407e-06,
+      "loss": 2.4823,
+      "step": 4547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.242661128346466e-06,
+      "loss": 2.4202,
+      "step": 4548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238668113219349e-06,
+      "loss": 2.5571,
+      "step": 4549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234675387892037e-06,
+      "loss": 2.4722,
+      "step": 4550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23068295302147e-06,
+      "loss": 2.5769,
+      "step": 4551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226690809264546e-06,
+      "loss": 2.4773,
+      "step": 4552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.222698957278112e-06,
+      "loss": 2.419,
+      "step": 4553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.218707397718961e-06,
+      "loss": 2.4888,
+      "step": 4554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.214716131243843e-06,
+      "loss": 2.4392,
+      "step": 4555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210725158509457e-06,
+      "loss": 2.4117,
+      "step": 4556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.206734480172461e-06,
+      "loss": 2.4425,
+      "step": 4557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202744096889455e-06,
+      "loss": 2.4819,
+      "step": 4558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.198754009316996e-06,
+      "loss": 2.5615,
+      "step": 4559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194764218111587e-06,
+      "loss": 2.5549,
+      "step": 4560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.190774723929694e-06,
+      "loss": 2.5531,
+      "step": 4561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186785527427721e-06,
+      "loss": 2.4665,
+      "step": 4562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18279662926203e-06,
+      "loss": 2.5245,
+      "step": 4563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178808030088927e-06,
+      "loss": 2.4922,
+      "step": 4564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.174819730564682e-06,
+      "loss": 2.4493,
+      "step": 4565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.170831731345507e-06,
+      "loss": 2.4438,
+      "step": 4566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.166844033087561e-06,
+      "loss": 2.4514,
+      "step": 4567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.162856636446956e-06,
+      "loss": 2.5395,
+      "step": 4568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15886954207976e-06,
+      "loss": 2.4502,
+      "step": 4569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.154882750641993e-06,
+      "loss": 2.4483,
+      "step": 4570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15089626278961e-06,
+      "loss": 2.6233,
+      "step": 4571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14691007917853e-06,
+      "loss": 2.4768,
+      "step": 4572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142924200464615e-06,
+      "loss": 2.5569,
+      "step": 4573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138938627303685e-06,
+      "loss": 2.5025,
+      "step": 4574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134953360351498e-06,
+      "loss": 2.5569,
+      "step": 4575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.130968400263772e-06,
+      "loss": 2.4557,
+      "step": 4576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126983747696166e-06,
+      "loss": 2.556,
+      "step": 4577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122999403304296e-06,
+      "loss": 2.4171,
+      "step": 4578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.119015367743727e-06,
+      "loss": 2.4443,
+      "step": 4579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115031641669966e-06,
+      "loss": 2.4256,
+      "step": 4580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.111048225738469e-06,
+      "loss": 2.5295,
+      "step": 4581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.107065120604654e-06,
+      "loss": 2.5308,
+      "step": 4582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103082326923875e-06,
+      "loss": 2.3865,
+      "step": 4583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099099845351438e-06,
+      "loss": 2.5389,
+      "step": 4584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.095117676542599e-06,
+      "loss": 2.4247,
+      "step": 4585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.091135821152564e-06,
+      "loss": 2.5193,
+      "step": 4586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.087154279836484e-06,
+      "loss": 2.4832,
+      "step": 4587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.083173053249458e-06,
+      "loss": 2.5142,
+      "step": 4588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.079192142046536e-06,
+      "loss": 2.3964,
+      "step": 4589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07521154688272e-06,
+      "loss": 2.4227,
+      "step": 4590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.071231268412952e-06,
+      "loss": 2.5027,
+      "step": 4591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.067251307292119e-06,
+      "loss": 2.436,
+      "step": 4592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063271664175072e-06,
+      "loss": 2.4456,
+      "step": 4593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.059292339716586e-06,
+      "loss": 2.4955,
+      "step": 4594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05531333457141e-06,
+      "loss": 2.5065,
+      "step": 4595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051334649394226e-06,
+      "loss": 2.5403,
+      "step": 4596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04735628483966e-06,
+      "loss": 2.4071,
+      "step": 4597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.043378241562289e-06,
+      "loss": 2.4697,
+      "step": 4598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.039400520216641e-06,
+      "loss": 2.5311,
+      "step": 4599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.035423121457192e-06,
+      "loss": 2.5058,
+      "step": 4600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.031446045938354e-06,
+      "loss": 2.4712,
+      "step": 4601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.027469294314499e-06,
+      "loss": 2.5298,
+      "step": 4602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023492867239935e-06,
+      "loss": 2.4552,
+      "step": 4603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.019516765368926e-06,
+      "loss": 2.4513,
+      "step": 4604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.015540989355674e-06,
+      "loss": 2.4693,
+      "step": 4605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01156553985433e-06,
+      "loss": 2.452,
+      "step": 4606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007590417519e-06,
+      "loss": 2.5381,
+      "step": 4607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.003615623003722e-06,
+      "loss": 2.4316,
+      "step": 4608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999641156962487e-06,
+      "loss": 2.5671,
+      "step": 4609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.995667020049236e-06,
+      "loss": 2.5349,
+      "step": 4610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.991693212917843e-06,
+      "loss": 2.4919,
+      "step": 4611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987719736222146e-06,
+      "loss": 2.5338,
+      "step": 4612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.983746590615913e-06,
+      "loss": 2.5373,
+      "step": 4613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.979773776752867e-06,
+      "loss": 2.429,
+      "step": 4614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.975801295286668e-06,
+      "loss": 2.5768,
+      "step": 4615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97182914687093e-06,
+      "loss": 2.4705,
+      "step": 4616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96785733215921e-06,
+      "loss": 2.459,
+      "step": 4617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.963885851805003e-06,
+      "loss": 2.4549,
+      "step": 4618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.959914706461756e-06,
+      "loss": 2.4227,
+      "step": 4619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.955943896782864e-06,
+      "loss": 2.5531,
+      "step": 4620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.951973423421659e-06,
+      "loss": 2.4807,
+      "step": 4621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948003287031418e-06,
+      "loss": 2.4432,
+      "step": 4622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944033488265365e-06,
+      "loss": 2.526,
+      "step": 4623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940064027776675e-06,
+      "loss": 2.4712,
+      "step": 4624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.936094906218458e-06,
+      "loss": 2.4046,
+      "step": 4625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932126124243768e-06,
+      "loss": 2.4648,
+      "step": 4626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.928157682505609e-06,
+      "loss": 2.4241,
+      "step": 4627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92418958165693e-06,
+      "loss": 2.5369,
+      "step": 4628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.920221822350614e-06,
+      "loss": 2.3435,
+      "step": 4629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9162544052395e-06,
+      "loss": 2.5232,
+      "step": 4630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91228733097636e-06,
+      "loss": 2.5693,
+      "step": 4631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908320600213914e-06,
+      "loss": 2.4944,
+      "step": 4632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90435421360483e-06,
+      "loss": 2.471,
+      "step": 4633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.900388171801718e-06,
+      "loss": 2.4284,
+      "step": 4634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89642247545712e-06,
+      "loss": 2.5276,
+      "step": 4635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892457125223532e-06,
+      "loss": 2.4696,
+      "step": 4636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.888492121753397e-06,
+      "loss": 2.49,
+      "step": 4637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884527465699092e-06,
+      "loss": 2.5565,
+      "step": 4638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.880563157712935e-06,
+      "loss": 2.428,
+      "step": 4639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876599198447191e-06,
+      "loss": 2.4979,
+      "step": 4640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872635588554076e-06,
+      "loss": 2.4439,
+      "step": 4641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868672328685738e-06,
+      "loss": 2.4897,
+      "step": 4642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.864709419494264e-06,
+      "loss": 2.5227,
+      "step": 4643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86074686163169e-06,
+      "loss": 2.5064,
+      "step": 4644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856784655750003e-06,
+      "loss": 2.4726,
+      "step": 4645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852822802501112e-06,
+      "loss": 2.5331,
+      "step": 4646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.848861302536884e-06,
+      "loss": 2.3985,
+      "step": 4647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.844900156509117e-06,
+      "loss": 2.4152,
+      "step": 4648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.840939365069562e-06,
+      "loss": 2.4754,
+      "step": 4649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836978928869902e-06,
+      "loss": 2.3806,
+      "step": 4650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83301884856177e-06,
+      "loss": 2.5831,
+      "step": 4651
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829059124796729e-06,
+      "loss": 2.4061,
+      "step": 4652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82509975822629e-06,
+      "loss": 2.4905,
+      "step": 4653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.821140749501913e-06,
+      "loss": 2.352,
+      "step": 4654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.817182099274988e-06,
+      "loss": 2.4794,
+      "step": 4655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.813223808196845e-06,
+      "loss": 2.4299,
+      "step": 4656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80926587691876e-06,
+      "loss": 2.419,
+      "step": 4657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805308306091959e-06,
+      "loss": 2.5052,
+      "step": 4658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.801351096367588e-06,
+      "loss": 2.3621,
+      "step": 4659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79739424839675e-06,
+      "loss": 2.5494,
+      "step": 4660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.793437762830474e-06,
+      "loss": 2.4965,
+      "step": 4661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789481640319752e-06,
+      "loss": 2.5458,
+      "step": 4662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.785525881515493e-06,
+      "loss": 2.5887,
+      "step": 4663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.781570487068561e-06,
+      "loss": 2.4742,
+      "step": 4664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777615457629747e-06,
+      "loss": 2.5908,
+      "step": 4665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.7736607938498e-06,
+      "loss": 2.5356,
+      "step": 4666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76970649637939e-06,
+      "loss": 2.5306,
+      "step": 4667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76575256586914e-06,
+      "loss": 2.5085,
+      "step": 4668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761799002969606e-06,
+      "loss": 2.556,
+      "step": 4669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.757845808331282e-06,
+      "loss": 2.4974,
+      "step": 4670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.753892982604613e-06,
+      "loss": 2.3473,
+      "step": 4671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74994052643997e-06,
+      "loss": 2.45,
+      "step": 4672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745988440487667e-06,
+      "loss": 2.4513,
+      "step": 4673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742036725397959e-06,
+      "loss": 2.4271,
+      "step": 4674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.738085381821045e-06,
+      "loss": 2.4521,
+      "step": 4675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.734134410407049e-06,
+      "loss": 2.5239,
+      "step": 4676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73018381180605e-06,
+      "loss": 2.5192,
+      "step": 4677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.726233586668046e-06,
+      "loss": 2.5442,
+      "step": 4678
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.722283735643e-06,
+      "loss": 2.5659,
+      "step": 4679
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.718334259380787e-06,
+      "loss": 2.5032,
+      "step": 4680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.714385158531238e-06,
+      "loss": 2.5065,
+      "step": 4681
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71043643374411e-06,
+      "loss": 2.3886,
+      "step": 4682
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706488085669114e-06,
+      "loss": 2.4723,
+      "step": 4683
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.702540114955883e-06,
+      "loss": 2.4525,
+      "step": 4684
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.698592522253997e-06,
+      "loss": 2.4618,
+      "step": 4685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694645308212964e-06,
+      "loss": 2.4136,
+      "step": 4686
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.690698473482246e-06,
+      "loss": 2.4599,
+      "step": 4687
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68675201871123e-06,
+      "loss": 2.4636,
+      "step": 4688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.682805944549244e-06,
+      "loss": 2.4364,
+      "step": 4689
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.678860251645551e-06,
+      "loss": 2.4872,
+      "step": 4690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.674914940649353e-06,
+      "loss": 2.3674,
+      "step": 4691
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670970012209795e-06,
+      "loss": 2.4975,
+      "step": 4692
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66702546697595e-06,
+      "loss": 2.438,
+      "step": 4693
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.663081305596832e-06,
+      "loss": 2.4489,
+      "step": 4694
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659137528721388e-06,
+      "loss": 2.4603,
+      "step": 4695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655194136998515e-06,
+      "loss": 2.4991,
+      "step": 4696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651251131077027e-06,
+      "loss": 2.499,
+      "step": 4697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647308511605692e-06,
+      "loss": 2.5046,
+      "step": 4698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643366279233196e-06,
+      "loss": 2.283,
+      "step": 4699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.639424434608184e-06,
+      "loss": 2.423,
+      "step": 4700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635482978379218e-06,
+      "loss": 2.4743,
+      "step": 4701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.631541911194809e-06,
+      "loss": 2.4383,
+      "step": 4702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.627601233703391e-06,
+      "loss": 2.5014,
+      "step": 4703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623660946553347e-06,
+      "loss": 2.416,
+      "step": 4704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.619721050392991e-06,
+      "loss": 2.409,
+      "step": 4705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.615781545870569e-06,
+      "loss": 2.5202,
+      "step": 4706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61184243363426e-06,
+      "loss": 2.4908,
+      "step": 4707
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607903714332193e-06,
+      "loss": 2.4489,
+      "step": 4708
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.603965388612421e-06,
+      "loss": 2.5007,
+      "step": 4709
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.600027457122931e-06,
+      "loss": 2.5,
+      "step": 4710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.596089920511651e-06,
+      "loss": 2.5133,
+      "step": 4711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.592152779426437e-06,
+      "loss": 2.4615,
+      "step": 4712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5882160345150926e-06,
+      "loss": 2.4167,
+      "step": 4713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5842796864253424e-06,
+      "loss": 2.4431,
+      "step": 4714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.580343735804852e-06,
+      "loss": 2.5021,
+      "step": 4715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.576408183301218e-06,
+      "loss": 2.4933,
+      "step": 4716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572473029561983e-06,
+      "loss": 2.5114,
+      "step": 4717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.568538275234608e-06,
+      "loss": 2.4257,
+      "step": 4718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5646039209665e-06,
+      "loss": 2.3826,
+      "step": 4719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.560669967404989e-06,
+      "loss": 2.5086,
+      "step": 4720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556736415197353e-06,
+      "loss": 2.4752,
+      "step": 4721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.552803264990798e-06,
+      "loss": 2.4631,
+      "step": 4722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.548870517432458e-06,
+      "loss": 2.5003,
+      "step": 4723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.544938173169404e-06,
+      "loss": 2.5316,
+      "step": 4724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541006232848647e-06,
+      "loss": 2.4876,
+      "step": 4725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.537074697117127e-06,
+      "loss": 2.4588,
+      "step": 4726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5331435666217135e-06,
+      "loss": 2.5314,
+      "step": 4727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.529212842009216e-06,
+      "loss": 2.4763,
+      "step": 4728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525282523926369e-06,
+      "loss": 2.4412,
+      "step": 4729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.521352613019852e-06,
+      "loss": 2.5105,
+      "step": 4730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.517423109936267e-06,
+      "loss": 2.4595,
+      "step": 4731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5134940153221556e-06,
+      "loss": 2.5623,
+      "step": 4732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5095653298239835e-06,
+      "loss": 2.4952,
+      "step": 4733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.50563705408816e-06,
+      "loss": 2.4977,
+      "step": 4734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.501709188761023e-06,
+      "loss": 2.526,
+      "step": 4735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4977817344888384e-06,
+      "loss": 2.5213,
+      "step": 4736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4938546919178055e-06,
+      "loss": 2.5591,
+      "step": 4737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489928061694063e-06,
+      "loss": 2.5133,
+      "step": 4738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486001844463677e-06,
+      "loss": 2.4999,
+      "step": 4739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.482076040872646e-06,
+      "loss": 2.4544,
+      "step": 4740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478150651566891e-06,
+      "loss": 2.3779,
+      "step": 4741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4742256771922835e-06,
+      "loss": 2.4072,
+      "step": 4742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470301118394616e-06,
+      "loss": 2.3494,
+      "step": 4743
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.466376975819611e-06,
+      "loss": 2.4254,
+      "step": 4744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462453250112924e-06,
+      "loss": 2.519,
+      "step": 4745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.458529941920147e-06,
+      "loss": 2.4277,
+      "step": 4746
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.454607051886801e-06,
+      "loss": 2.3291,
+      "step": 4747
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.450684580658329e-06,
+      "loss": 2.4803,
+      "step": 4748
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446762528880121e-06,
+      "loss": 2.4578,
+      "step": 4749
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.442840897197481e-06,
+      "loss": 2.3827,
+      "step": 4750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.438919686255659e-06,
+      "loss": 2.5329,
+      "step": 4751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43499889669983e-06,
+      "loss": 2.5284,
+      "step": 4752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.431078529175097e-06,
+      "loss": 2.489,
+      "step": 4753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.42715858432649e-06,
+      "loss": 2.3893,
+      "step": 4754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.423239062798983e-06,
+      "loss": 2.4128,
+      "step": 4755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.419319965237474e-06,
+      "loss": 2.4283,
+      "step": 4756
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4154012922867835e-06,
+      "loss": 2.4949,
+      "step": 4757
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.411483044591666e-06,
+      "loss": 2.4315,
+      "step": 4758
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.407565222796815e-06,
+      "loss": 2.508,
+      "step": 4759
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.403647827546847e-06,
+      "loss": 2.5247,
+      "step": 4760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399730859486305e-06,
+      "loss": 2.5416,
+      "step": 4761
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.395814319259662e-06,
+      "loss": 2.4777,
+      "step": 4762
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.391898207511335e-06,
+      "loss": 2.5133,
+      "step": 4763
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.387982524885652e-06,
+      "loss": 2.4874,
+      "step": 4764
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3840672720268755e-06,
+      "loss": 2.5191,
+      "step": 4765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3801524495792035e-06,
+      "loss": 2.4539,
+      "step": 4766
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.37623805818676e-06,
+      "loss": 2.45,
+      "step": 4767
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372324098493593e-06,
+      "loss": 2.5285,
+      "step": 4768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.368410571143691e-06,
+      "loss": 2.3058,
+      "step": 4769
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.364497476780958e-06,
+      "loss": 2.4256,
+      "step": 4770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.36058481604923e-06,
+      "loss": 2.4635,
+      "step": 4771
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356672589592282e-06,
+      "loss": 2.443,
+      "step": 4772
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.352760798053808e-06,
+      "loss": 2.4084,
+      "step": 4773
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.348849442077433e-06,
+      "loss": 2.507,
+      "step": 4774
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344938522306703e-06,
+      "loss": 2.4505,
+      "step": 4775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34102803938511e-06,
+      "loss": 2.5384,
+      "step": 4776
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337117993956058e-06,
+      "loss": 2.5274,
+      "step": 4777
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.333208386662881e-06,
+      "loss": 2.526,
+      "step": 4778
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.329299218148846e-06,
+      "loss": 2.3644,
+      "step": 4779
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.325390489057152e-06,
+      "loss": 2.5111,
+      "step": 4780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.321482200030914e-06,
+      "loss": 2.5278,
+      "step": 4781
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.31757435171318e-06,
+      "loss": 2.4647,
+      "step": 4782
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.313666944746924e-06,
+      "loss": 2.483,
+      "step": 4783
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.309759979775058e-06,
+      "loss": 2.5162,
+      "step": 4784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.305853457440404e-06,
+      "loss": 2.4679,
+      "step": 4785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.301947378385723e-06,
+      "loss": 2.493,
+      "step": 4786
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2980417432537e-06,
+      "loss": 2.4998,
+      "step": 4787
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29413655268694e-06,
+      "loss": 2.6036,
+      "step": 4788
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29023180732799e-06,
+      "loss": 2.3849,
+      "step": 4789
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.286327507819313e-06,
+      "loss": 2.5155,
+      "step": 4790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.282423654803299e-06,
+      "loss": 2.48,
+      "step": 4791
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.278520248922267e-06,
+      "loss": 2.4477,
+      "step": 4792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.274617290818462e-06,
+      "loss": 2.5401,
+      "step": 4793
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.270714781134056e-06,
+      "loss": 2.5056,
+      "step": 4794
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.266812720511145e-06,
+      "loss": 2.4704,
+      "step": 4795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.262911109591751e-06,
+      "loss": 2.4807,
+      "step": 4796
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.259009949017828e-06,
+      "loss": 2.4455,
+      "step": 4797
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25510923943125e-06,
+      "loss": 2.4582,
+      "step": 4798
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251208981473816e-06,
+      "loss": 2.4891,
+      "step": 4799
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.247309175787252e-06,
+      "loss": 2.393,
+      "step": 4800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.243409823013214e-06,
+      "loss": 2.4563,
+      "step": 4801
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.239510923793278e-06,
+      "loss": 2.5273,
+      "step": 4802
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23561247876895e-06,
+      "loss": 2.379,
+      "step": 4803
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.231714488581651e-06,
+      "loss": 2.4664,
+      "step": 4804
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227816953872746e-06,
+      "loss": 2.4832,
+      "step": 4805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.223919875283505e-06,
+      "loss": 2.474,
+      "step": 4806
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.220023253455135e-06,
+      "loss": 2.4048,
+      "step": 4807
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216127089028764e-06,
+      "loss": 2.4467,
+      "step": 4808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.212231382645445e-06,
+      "loss": 2.524,
+      "step": 4809
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.208336134946157e-06,
+      "loss": 2.4417,
+      "step": 4810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204441346571805e-06,
+      "loss": 2.427,
+      "step": 4811
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.200547018163209e-06,
+      "loss": 2.4951,
+      "step": 4812
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.196653150361121e-06,
+      "loss": 2.4318,
+      "step": 4813
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.192759743806226e-06,
+      "loss": 2.4454,
+      "step": 4814
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.188866799139118e-06,
+      "loss": 2.4224,
+      "step": 4815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1849743170003165e-06,
+      "loss": 2.6046,
+      "step": 4816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.18108229803027e-06,
+      "loss": 2.5019,
+      "step": 4817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177190742869357e-06,
+      "loss": 2.4203,
+      "step": 4818
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1732996521578645e-06,
+      "loss": 2.3986,
+      "step": 4819
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1694090265360175e-06,
+      "loss": 2.5082,
+      "step": 4820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.16551886664395e-06,
+      "loss": 2.5272,
+      "step": 4821
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.161629173121736e-06,
+      "loss": 2.5172,
+      "step": 4822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.157739946609359e-06,
+      "loss": 2.5682,
+      "step": 4823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153851187746732e-06,
+      "loss": 2.4382,
+      "step": 4824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.14996289717369e-06,
+      "loss": 2.4023,
+      "step": 4825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.146075075529988e-06,
+      "loss": 2.4205,
+      "step": 4826
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.142187723455314e-06,
+      "loss": 2.4808,
+      "step": 4827
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1383008415892655e-06,
+      "loss": 2.4025,
+      "step": 4828
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1344144305713685e-06,
+      "loss": 2.4246,
+      "step": 4829
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.130528491041071e-06,
+      "loss": 2.4749,
+      "step": 4830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.12664302363775e-06,
+      "loss": 2.5212,
+      "step": 4831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.122758029000696e-06,
+      "loss": 2.5334,
+      "step": 4832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.11887350776912e-06,
+      "loss": 2.5719,
+      "step": 4833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.11498946058216e-06,
+      "loss": 2.5014,
+      "step": 4834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.111105888078884e-06,
+      "loss": 2.5005,
+      "step": 4835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.107222790898265e-06,
+      "loss": 2.5079,
+      "step": 4836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1033401696792134e-06,
+      "loss": 2.5023,
+      "step": 4837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0994580250605434e-06,
+      "loss": 2.6153,
+      "step": 4838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.095576357681015e-06,
+      "loss": 2.4391,
+      "step": 4839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.091695168179287e-06,
+      "loss": 2.4907,
+      "step": 4840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087814457193953e-06,
+      "loss": 2.4167,
+      "step": 4841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.08393422536352e-06,
+      "loss": 2.4691,
+      "step": 4842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.080054473326423e-06,
+      "loss": 2.4556,
+      "step": 4843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076175201721016e-06,
+      "loss": 2.5684,
+      "step": 4844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.07229641118557e-06,
+      "loss": 2.5163,
+      "step": 4845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.068418102358281e-06,
+      "loss": 2.4962,
+      "step": 4846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.064540275877262e-06,
+      "loss": 2.4976,
+      "step": 4847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.060662932380556e-06,
+      "loss": 2.5068,
+      "step": 4848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.056786072506112e-06,
+      "loss": 2.4913,
+      "step": 4849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.052909696891811e-06,
+      "loss": 2.48,
+      "step": 4850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.049033806175447e-06,
+      "loss": 2.4737,
+      "step": 4851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0451584009947426e-06,
+      "loss": 2.4549,
+      "step": 4852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.041283481987333e-06,
+      "loss": 2.3829,
+      "step": 4853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037409049790778e-06,
+      "loss": 2.4059,
+      "step": 4854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0335351050425484e-06,
+      "loss": 2.4479,
+      "step": 4855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0296616483800506e-06,
+      "loss": 2.4293,
+      "step": 4856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.025788680440596e-06,
+      "loss": 2.4697,
+      "step": 4857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.021916201861425e-06,
+      "loss": 2.5314,
+      "step": 4858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0180442132796875e-06,
+      "loss": 2.3657,
+      "step": 4859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0141727153324654e-06,
+      "loss": 2.5399,
+      "step": 4860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.010301708656755e-06,
+      "loss": 2.4027,
+      "step": 4861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.006431193889465e-06,
+      "loss": 2.4342,
+      "step": 4862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.002561171667428e-06,
+      "loss": 2.5621,
+      "step": 4863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9986916426274e-06,
+      "loss": 2.5023,
+      "step": 4864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.994822607406051e-06,
+      "loss": 2.4854,
+      "step": 4865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.99095406663997e-06,
+      "loss": 2.4257,
+      "step": 4866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.987086020965665e-06,
+      "loss": 2.4432,
+      "step": 4867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9832184710195615e-06,
+      "loss": 2.5031,
+      "step": 4868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.979351417438009e-06,
+      "loss": 2.4241,
+      "step": 4869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.975484860857268e-06,
+      "loss": 2.4224,
+      "step": 4870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971618801913522e-06,
+      "loss": 2.4132,
+      "step": 4871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967753241242865e-06,
+      "loss": 2.4796,
+      "step": 4872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.963888179481325e-06,
+      "loss": 2.413,
+      "step": 4873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.960023617264831e-06,
+      "loss": 2.3283,
+      "step": 4874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956159555229238e-06,
+      "loss": 2.4044,
+      "step": 4875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.952295994010316e-06,
+      "loss": 2.5222,
+      "step": 4876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.948432934243757e-06,
+      "loss": 2.5248,
+      "step": 4877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.94457037656517e-06,
+      "loss": 2.5563,
+      "step": 4878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.940708321610072e-06,
+      "loss": 2.5082,
+      "step": 4879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.936846770013905e-06,
+      "loss": 2.5167,
+      "step": 4880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9329857224120324e-06,
+      "loss": 2.4919,
+      "step": 4881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929125179439729e-06,
+      "loss": 2.4865,
+      "step": 4882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.925265141732181e-06,
+      "loss": 2.5031,
+      "step": 4883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921405609924505e-06,
+      "loss": 2.5063,
+      "step": 4884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.91754658465172e-06,
+      "loss": 2.4375,
+      "step": 4885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.913688066548776e-06,
+      "loss": 2.5471,
+      "step": 4886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 2.4185,
+      "step": 4887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9059725543917525e-06,
+      "loss": 2.4625,
+      "step": 4888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9021155616071375e-06,
+      "loss": 2.6135,
+      "step": 4889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.898259078531299e-06,
+      "loss": 2.428,
+      "step": 4890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894403105798757e-06,
+      "loss": 2.3771,
+      "step": 4891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890547644043953e-06,
+      "loss": 2.4652,
+      "step": 4892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88669269390124e-06,
+      "loss": 2.3827,
+      "step": 4893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.882838256004894e-06,
+      "loss": 2.5193,
+      "step": 4894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.878984330989104e-06,
+      "loss": 2.4688,
+      "step": 4895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.875130919487971e-06,
+      "loss": 2.3886,
+      "step": 4896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8712780221355124e-06,
+      "loss": 2.4769,
+      "step": 4897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.867425639565664e-06,
+      "loss": 2.5023,
+      "step": 4898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86357377241228e-06,
+      "loss": 2.5029,
+      "step": 4899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.859722421309118e-06,
+      "loss": 2.4872,
+      "step": 4900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8558715868898606e-06,
+      "loss": 2.4464,
+      "step": 4901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.852021269788107e-06,
+      "loss": 2.4398,
+      "step": 4902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8481714706373635e-06,
+      "loss": 2.3907,
+      "step": 4903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.844322190071054e-06,
+      "loss": 2.5646,
+      "step": 4904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8404734287225205e-06,
+      "loss": 2.5246,
+      "step": 4905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.836625187225011e-06,
+      "loss": 2.4662,
+      "step": 4906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832777466211702e-06,
+      "loss": 2.4632,
+      "step": 4907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.828930266315671e-06,
+      "loss": 2.519,
+      "step": 4908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8250835881699185e-06,
+      "loss": 2.4106,
+      "step": 4909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.821237432407349e-06,
+      "loss": 2.4879,
+      "step": 4910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.817391799660793e-06,
+      "loss": 2.4863,
+      "step": 4911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.813546690562993e-06,
+      "loss": 2.4384,
+      "step": 4912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8097021057465966e-06,
+      "loss": 2.4726,
+      "step": 4913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805858045844167e-06,
+      "loss": 2.5154,
+      "step": 4914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8020145114881934e-06,
+      "loss": 2.4054,
+      "step": 4915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.798171503311066e-06,
+      "loss": 2.4583,
+      "step": 4916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.794329021945089e-06,
+      "loss": 2.529,
+      "step": 4917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790487068022485e-06,
+      "loss": 2.4833,
+      "step": 4918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786645642175393e-06,
+      "loss": 2.4853,
+      "step": 4919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782804745035856e-06,
+      "loss": 2.4571,
+      "step": 4920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7789643772358306e-06,
+      "loss": 2.5542,
+      "step": 4921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.77512453940719e-06,
+      "loss": 2.4827,
+      "step": 4922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.771285232181727e-06,
+      "loss": 2.47,
+      "step": 4923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767446456191136e-06,
+      "loss": 2.4572,
+      "step": 4924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7636082120670245e-06,
+      "loss": 2.516,
+      "step": 4925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.759770500440921e-06,
+      "loss": 2.5108,
+      "step": 4926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.755933321944255e-06,
+      "loss": 2.4994,
+      "step": 4927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752096677208382e-06,
+      "loss": 2.5302,
+      "step": 4928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.74826056686456e-06,
+      "loss": 2.5116,
+      "step": 4929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744424991543962e-06,
+      "loss": 2.4927,
+      "step": 4930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740589951877665e-06,
+      "loss": 2.4701,
+      "step": 4931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7367554484966745e-06,
+      "loss": 2.4686,
+      "step": 4932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.732921482031898e-06,
+      "loss": 2.436,
+      "step": 4933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729088053114148e-06,
+      "loss": 2.5526,
+      "step": 4934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.725255162374158e-06,
+      "loss": 2.5535,
+      "step": 4935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.721422810442578e-06,
+      "loss": 2.5395,
+      "step": 4936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.717590997949957e-06,
+      "loss": 2.4999,
+      "step": 4937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7137597255267566e-06,
+      "loss": 2.4562,
+      "step": 4938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.709928993803354e-06,
+      "loss": 2.4821,
+      "step": 4939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.706098803410044e-06,
+      "loss": 2.4601,
+      "step": 4940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.702269154977017e-06,
+      "loss": 2.4847,
+      "step": 4941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6984400491343886e-06,
+      "loss": 2.4458,
+      "step": 4942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.694611486512174e-06,
+      "loss": 2.577,
+      "step": 4943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690783467740301e-06,
+      "loss": 2.3865,
+      "step": 4944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.686955993448617e-06,
+      "loss": 2.5122,
+      "step": 4945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.683129064266873e-06,
+      "loss": 2.4521,
+      "step": 4946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679302680824728e-06,
+      "loss": 2.3971,
+      "step": 4947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.675476843751752e-06,
+      "loss": 2.411,
+      "step": 4948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.671651553677431e-06,
+      "loss": 2.526,
+      "step": 4949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.667826811231157e-06,
+      "loss": 2.5297,
+      "step": 4950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664002617042231e-06,
+      "loss": 2.6335,
+      "step": 4951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6601789717398615e-06,
+      "loss": 2.5071,
+      "step": 4952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.656355875953178e-06,
+      "loss": 2.4953,
+      "step": 4953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652533330311205e-06,
+      "loss": 2.4719,
+      "step": 4954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6487113354428836e-06,
+      "loss": 2.5167,
+      "step": 4955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.644889891977062e-06,
+      "loss": 2.4791,
+      "step": 4956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.641069000542506e-06,
+      "loss": 2.4455,
+      "step": 4957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6372486617678785e-06,
+      "loss": 2.437,
+      "step": 4958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.633428876281762e-06,
+      "loss": 2.5095,
+      "step": 4959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629609644712633e-06,
+      "loss": 2.4721,
+      "step": 4960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.625790967688899e-06,
+      "loss": 2.5138,
+      "step": 4961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.621972845838856e-06,
+      "loss": 2.3922,
+      "step": 4962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618155279790719e-06,
+      "loss": 2.4418,
+      "step": 4963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614338270172609e-06,
+      "loss": 2.4844,
+      "step": 4964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.610521817612553e-06,
+      "loss": 2.5054,
+      "step": 4965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606705922738492e-06,
+      "loss": 2.4317,
+      "step": 4966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.602890586178275e-06,
+      "loss": 2.4629,
+      "step": 4967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599075808559652e-06,
+      "loss": 2.4683,
+      "step": 4968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.595261590510283e-06,
+      "loss": 2.4335,
+      "step": 4969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591447932657746e-06,
+      "loss": 2.4318,
+      "step": 4970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.587634835629517e-06,
+      "loss": 2.5395,
+      "step": 4971
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.583822300052976e-06,
+      "loss": 2.402,
+      "step": 4972
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.580010326555419e-06,
+      "loss": 2.4737,
+      "step": 4973
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.576198915764054e-06,
+      "loss": 2.3512,
+      "step": 4974
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.572388068305981e-06,
+      "loss": 2.5655,
+      "step": 4975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.568577784808221e-06,
+      "loss": 2.4407,
+      "step": 4976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56476806589769e-06,
+      "loss": 2.5398,
+      "step": 4977
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5609589122012275e-06,
+      "loss": 2.4927,
+      "step": 4978
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5571503243455635e-06,
+      "loss": 2.4794,
+      "step": 4979
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.553342302957346e-06,
+      "loss": 2.5485,
+      "step": 4980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549534848663126e-06,
+      "loss": 2.3625,
+      "step": 4981
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545727962089355e-06,
+      "loss": 2.3457,
+      "step": 4982
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541921643862401e-06,
+      "loss": 2.5367,
+      "step": 4983
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5381158946085385e-06,
+      "loss": 2.5778,
+      "step": 4984
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.534310714953938e-06,
+      "loss": 2.5241,
+      "step": 4985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5305061055246836e-06,
+      "loss": 2.5003,
+      "step": 4986
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5267020669467704e-06,
+      "loss": 2.376,
+      "step": 4987
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5228985998460906e-06,
+      "loss": 2.3872,
+      "step": 4988
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519095704848444e-06,
+      "loss": 2.5374,
+      "step": 4989
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515293382579535e-06,
+      "loss": 2.4809,
+      "step": 4990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.511491633664986e-06,
+      "loss": 2.55,
+      "step": 4991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507690458730307e-06,
+      "loss": 2.3813,
+      "step": 4992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503889858400928e-06,
+      "loss": 2.4479,
+      "step": 4993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500089833302172e-06,
+      "loss": 2.55,
+      "step": 4994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.496290384059284e-06,
+      "loss": 2.5485,
+      "step": 4995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.492491511297396e-06,
+      "loss": 2.5085,
+      "step": 4996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4886932156415596e-06,
+      "loss": 2.4451,
+      "step": 4997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4848954977167165e-06,
+      "loss": 2.3805,
+      "step": 4998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4810983581477326e-06,
+      "loss": 2.4233,
+      "step": 4999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.477301797559361e-06,
+      "loss": 2.4677,
+      "step": 5000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473505816576273e-06,
+      "loss": 2.4862,
+      "step": 5001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.469710415823032e-06,
+      "loss": 2.4769,
+      "step": 5002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.465915595924111e-06,
+      "loss": 2.4524,
+      "step": 5003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.462121357503899e-06,
+      "loss": 2.5473,
+      "step": 5004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458327701186671e-06,
+      "loss": 2.4719,
+      "step": 5005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.454534627596614e-06,
+      "loss": 2.5137,
+      "step": 5006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4507421373578195e-06,
+      "loss": 2.4167,
+      "step": 5007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.446950231094289e-06,
+      "loss": 2.4188,
+      "step": 5008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4431589094299144e-06,
+      "loss": 2.4422,
+      "step": 5009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439368172988505e-06,
+      "loss": 2.4121,
+      "step": 5010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4355780223937585e-06,
+      "loss": 2.46,
+      "step": 5011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.431788458269297e-06,
+      "loss": 2.4793,
+      "step": 5012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.427999481238625e-06,
+      "loss": 2.5008,
+      "step": 5013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.424211091925166e-06,
+      "loss": 2.485,
+      "step": 5014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4204232909522336e-06,
+      "loss": 2.4868,
+      "step": 5015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.416636078943058e-06,
+      "loss": 2.5835,
+      "step": 5016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4128494565207665e-06,
+      "loss": 2.4886,
+      "step": 5017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409063424308387e-06,
+      "loss": 2.4325,
+      "step": 5018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.405277982928848e-06,
+      "loss": 2.4349,
+      "step": 5019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.40149313300499e-06,
+      "loss": 2.3912,
+      "step": 5020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.397708875159554e-06,
+      "loss": 2.4483,
+      "step": 5021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393925210015175e-06,
+      "loss": 2.4957,
+      "step": 5022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3901421381943985e-06,
+      "loss": 2.3725,
+      "step": 5023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386359660319668e-06,
+      "loss": 2.3969,
+      "step": 5024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382577777013337e-06,
+      "loss": 2.4781,
+      "step": 5025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37879648889765e-06,
+      "loss": 2.5287,
+      "step": 5026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375015796594764e-06,
+      "loss": 2.5319,
+      "step": 5027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.371235700726727e-06,
+      "loss": 2.4254,
+      "step": 5028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.367456201915503e-06,
+      "loss": 2.535,
+      "step": 5029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.363677300782943e-06,
+      "loss": 2.5225,
+      "step": 5030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.359898997950812e-06,
+      "loss": 2.5465,
+      "step": 5031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.356121294040765e-06,
+      "loss": 2.5292,
+      "step": 5032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.352344189674367e-06,
+      "loss": 2.3825,
+      "step": 5033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3485676854730875e-06,
+      "loss": 2.5449,
+      "step": 5034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.344791782058286e-06,
+      "loss": 2.5175,
+      "step": 5035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.341016480051226e-06,
+      "loss": 2.4108,
+      "step": 5036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.337241780073078e-06,
+      "loss": 2.4545,
+      "step": 5037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333467682744914e-06,
+      "loss": 2.5051,
+      "step": 5038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3296941886877e-06,
+      "loss": 2.508,
+      "step": 5039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.325921298522304e-06,
+      "loss": 2.5077,
+      "step": 5040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.322149012869495e-06,
+      "loss": 2.3784,
+      "step": 5041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.318377332349952e-06,
+      "loss": 2.4813,
+      "step": 5042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.314606257584238e-06,
+      "loss": 2.4907,
+      "step": 5043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31083578919283e-06,
+      "loss": 2.4797,
+      "step": 5044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3070659277960965e-06,
+      "loss": 2.4698,
+      "step": 5045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.303296674014314e-06,
+      "loss": 2.5003,
+      "step": 5046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.299528028467652e-06,
+      "loss": 2.5769,
+      "step": 5047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295759991776182e-06,
+      "loss": 2.4831,
+      "step": 5048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291992564559874e-06,
+      "loss": 2.455,
+      "step": 5049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.288225747438604e-06,
+      "loss": 2.5196,
+      "step": 5050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.284459541032144e-06,
+      "loss": 2.5428,
+      "step": 5051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.280693945960161e-06,
+      "loss": 2.5446,
+      "step": 5052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.276928962842223e-06,
+      "loss": 2.4319,
+      "step": 5053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2731645922978064e-06,
+      "loss": 2.5461,
+      "step": 5054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.269400834946278e-06,
+      "loss": 2.4233,
+      "step": 5055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.265637691406903e-06,
+      "loss": 2.4614,
+      "step": 5056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.261875162298845e-06,
+      "loss": 2.4874,
+      "step": 5057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2581132482411774e-06,
+      "loss": 2.4294,
+      "step": 5058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254351949852864e-06,
+      "loss": 2.5023,
+      "step": 5059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.250591267752761e-06,
+      "loss": 2.4787,
+      "step": 5060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.246831202559638e-06,
+      "loss": 2.5128,
+      "step": 5061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243071754892148e-06,
+      "loss": 2.3407,
+      "step": 5062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.239312925368859e-06,
+      "loss": 2.5088,
+      "step": 5063
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235554714608219e-06,
+      "loss": 2.4634,
+      "step": 5064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.231797123228589e-06,
+      "loss": 2.5103,
+      "step": 5065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.228040151848217e-06,
+      "loss": 2.5098,
+      "step": 5066
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.224283801085259e-06,
+      "loss": 2.551,
+      "step": 5067
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.220528071557765e-06,
+      "loss": 2.4395,
+      "step": 5068
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.21677296388368e-06,
+      "loss": 2.4921,
+      "step": 5069
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.213018478680843e-06,
+      "loss": 2.5339,
+      "step": 5070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209264616567006e-06,
+      "loss": 2.4817,
+      "step": 5071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.205511378159804e-06,
+      "loss": 2.5399,
+      "step": 5072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.201758764076773e-06,
+      "loss": 2.4795,
+      "step": 5073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.19800677493535e-06,
+      "loss": 2.5207,
+      "step": 5074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.194255411352863e-06,
+      "loss": 2.4759,
+      "step": 5075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.190504673946547e-06,
+      "loss": 2.483,
+      "step": 5076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186754563333521e-06,
+      "loss": 2.4228,
+      "step": 5077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183005080130807e-06,
+      "loss": 2.4115,
+      "step": 5078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179256224955332e-06,
+      "loss": 2.5498,
+      "step": 5079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.175507998423908e-06,
+      "loss": 2.3948,
+      "step": 5080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171760401153242e-06,
+      "loss": 2.4337,
+      "step": 5081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168013433759952e-06,
+      "loss": 2.4449,
+      "step": 5082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164267096860534e-06,
+      "loss": 2.4701,
+      "step": 5083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1605213910713955e-06,
+      "loss": 2.4768,
+      "step": 5084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.156776317008836e-06,
+      "loss": 2.4938,
+      "step": 5085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1530318752890465e-06,
+      "loss": 2.5267,
+      "step": 5086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.149288066528112e-06,
+      "loss": 2.5022,
+      "step": 5087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.145544891342024e-06,
+      "loss": 2.4806,
+      "step": 5088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141802350346665e-06,
+      "loss": 2.4944,
+      "step": 5089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.138060444157807e-06,
+      "loss": 2.4716,
+      "step": 5090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.134319173391124e-06,
+      "loss": 2.479,
+      "step": 5091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.130578538662188e-06,
+      "loss": 2.5808,
+      "step": 5092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.126838540586462e-06,
+      "loss": 2.5453,
+      "step": 5093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123099179779297e-06,
+      "loss": 2.5085,
+      "step": 5094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1193604568559504e-06,
+      "loss": 2.5652,
+      "step": 5095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1156223724315775e-06,
+      "loss": 2.5143,
+      "step": 5096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1118849271212154e-06,
+      "loss": 2.5012,
+      "step": 5097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1081481215398035e-06,
+      "loss": 2.3897,
+      "step": 5098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.104411956302177e-06,
+      "loss": 2.5164,
+      "step": 5099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100676432023058e-06,
+      "loss": 2.39,
+      "step": 5100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.096941549317077e-06,
+      "loss": 2.4768,
+      "step": 5101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.093207308798747e-06,
+      "loss": 2.4222,
+      "step": 5102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.089473711082483e-06,
+      "loss": 2.4297,
+      "step": 5103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0857407567825814e-06,
+      "loss": 2.434,
+      "step": 5104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.082008446513251e-06,
+      "loss": 2.5466,
+      "step": 5105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.078276780888583e-06,
+      "loss": 2.5035,
+      "step": 5106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0745457605225634e-06,
+      "loss": 2.4005,
+      "step": 5107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.070815386029073e-06,
+      "loss": 2.5861,
+      "step": 5108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.067085658021892e-06,
+      "loss": 2.4718,
+      "step": 5109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0633565771146855e-06,
+      "loss": 2.494,
+      "step": 5110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059628143921015e-06,
+      "loss": 2.5184,
+      "step": 5111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.055900359054334e-06,
+      "loss": 2.5296,
+      "step": 5112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.052173223128e-06,
+      "loss": 2.5149,
+      "step": 5113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0484467367552515e-06,
+      "loss": 2.4707,
+      "step": 5114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0447209005492195e-06,
+      "loss": 2.5057,
+      "step": 5115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0409957151229325e-06,
+      "loss": 2.4837,
+      "step": 5116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0372711810893215e-06,
+      "loss": 2.4857,
+      "step": 5117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.033547299061191e-06,
+      "loss": 2.5169,
+      "step": 5118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.029824069651254e-06,
+      "loss": 2.5068,
+      "step": 5119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.026101493472107e-06,
+      "loss": 2.4808,
+      "step": 5120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.02237957113624e-06,
+      "loss": 2.4598,
+      "step": 5121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018658303256041e-06,
+      "loss": 2.4386,
+      "step": 5122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0149376904437884e-06,
+      "loss": 2.465,
+      "step": 5123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.011217733311646e-06,
+      "loss": 2.5109,
+      "step": 5124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.007498432471677e-06,
+      "loss": 2.5076,
+      "step": 5125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.003779788535837e-06,
+      "loss": 2.4465,
+      "step": 5126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.000061802115973e-06,
+      "loss": 2.4985,
+      "step": 5127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9963444738238144e-06,
+      "loss": 2.4813,
+      "step": 5128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.992627804270992e-06,
+      "loss": 2.5409,
+      "step": 5129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.988911794069031e-06,
+      "loss": 2.4762,
+      "step": 5130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985196443829341e-06,
+      "loss": 2.4903,
+      "step": 5131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.981481754163222e-06,
+      "loss": 2.3652,
+      "step": 5132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977767725681869e-06,
+      "loss": 2.4365,
+      "step": 5133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.974054358996373e-06,
+      "loss": 2.5002,
+      "step": 5134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.970341654717705e-06,
+      "loss": 2.4221,
+      "step": 5135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966629613456738e-06,
+      "loss": 2.4825,
+      "step": 5136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.962918235824227e-06,
+      "loss": 2.539,
+      "step": 5137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.959207522430819e-06,
+      "loss": 2.3934,
+      "step": 5138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955497473887058e-06,
+      "loss": 2.4263,
+      "step": 5139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.951788090803379e-06,
+      "loss": 2.4806,
+      "step": 5140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.948079373790094e-06,
+      "loss": 2.5383,
+      "step": 5141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.944371323457421e-06,
+      "loss": 2.6047,
+      "step": 5142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9406639404154635e-06,
+      "loss": 2.4809,
+      "step": 5143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936957225274209e-06,
+      "loss": 2.439,
+      "step": 5144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.933251178643542e-06,
+      "loss": 2.4637,
+      "step": 5145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.929545801133232e-06,
+      "loss": 2.4281,
+      "step": 5146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9258410933529484e-06,
+      "loss": 2.4976,
+      "step": 5147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922137055912238e-06,
+      "loss": 2.5186,
+      "step": 5148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.918433689420546e-06,
+      "loss": 2.4865,
+      "step": 5149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9147309944871954e-06,
+      "loss": 2.4333,
+      "step": 5150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.91102897172142e-06,
+      "loss": 2.4838,
+      "step": 5151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.90732762173232e-06,
+      "loss": 2.4241,
+      "step": 5152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903626945128899e-06,
+      "loss": 2.521,
+      "step": 5153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.899926942520044e-06,
+      "loss": 2.5141,
+      "step": 5154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.896227614514538e-06,
+      "loss": 2.4962,
+      "step": 5155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.892528961721041e-06,
+      "loss": 2.4969,
+      "step": 5156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888830984748115e-06,
+      "loss": 2.5199,
+      "step": 5157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.885133684204201e-06,
+      "loss": 2.5821,
+      "step": 5158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.88143706069763e-06,
+      "loss": 2.4696,
+      "step": 5159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.877741114836632e-06,
+      "loss": 2.5612,
+      "step": 5160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.874045847229311e-06,
+      "loss": 2.413,
+      "step": 5161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.870351258483667e-06,
+      "loss": 2.4178,
+      "step": 5162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866657349207588e-06,
+      "loss": 2.5301,
+      "step": 5163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.86296412000885e-06,
+      "loss": 2.4778,
+      "step": 5164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.859271571495116e-06,
+      "loss": 2.5489,
+      "step": 5165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.855579704273937e-06,
+      "loss": 2.5403,
+      "step": 5166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.851888518952749e-06,
+      "loss": 2.4967,
+      "step": 5167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848198016138889e-06,
+      "loss": 2.5016,
+      "step": 5168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.844508196439562e-06,
+      "loss": 2.4417,
+      "step": 5169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.840819060461878e-06,
+      "loss": 2.4025,
+      "step": 5170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.837130608812816e-06,
+      "loss": 2.3416,
+      "step": 5171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.83344284209927e-06,
+      "loss": 2.5392,
+      "step": 5172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.829755760927991e-06,
+      "loss": 2.5222,
+      "step": 5173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.826069365905635e-06,
+      "loss": 2.4558,
+      "step": 5174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8223836576387415e-06,
+      "loss": 2.4792,
+      "step": 5175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.818698636733738e-06,
+      "loss": 2.443,
+      "step": 5176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8150143037969345e-06,
+      "loss": 2.4766,
+      "step": 5177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.811330659434537e-06,
+      "loss": 2.5228,
+      "step": 5178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.807647704252624e-06,
+      "loss": 2.5226,
+      "step": 5179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.803965438857171e-06,
+      "loss": 2.4286,
+      "step": 5180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8002838638540385e-06,
+      "loss": 2.3957,
+      "step": 5181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.796602979848973e-06,
+      "loss": 2.5085,
+      "step": 5182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79292278744761e-06,
+      "loss": 2.5378,
+      "step": 5183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789243287255456e-06,
+      "loss": 2.4617,
+      "step": 5184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.785564479877927e-06,
+      "loss": 2.4404,
+      "step": 5185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7818863659203174e-06,
+      "loss": 2.5696,
+      "step": 5186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.778208945987791e-06,
+      "loss": 2.5524,
+      "step": 5187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.774532220685411e-06,
+      "loss": 2.4144,
+      "step": 5188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770856190618139e-06,
+      "loss": 2.4686,
+      "step": 5189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.767180856390797e-06,
+      "loss": 2.4741,
+      "step": 5190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.763506218608108e-06,
+      "loss": 2.4235,
+      "step": 5191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.759832277874674e-06,
+      "loss": 2.4544,
+      "step": 5192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756159034794988e-06,
+      "loss": 2.5047,
+      "step": 5193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.752486489973422e-06,
+      "loss": 2.4457,
+      "step": 5194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.748814644014245e-06,
+      "loss": 2.5097,
+      "step": 5195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7451434975215885e-06,
+      "loss": 2.4432,
+      "step": 5196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.74147305109949e-06,
+      "loss": 2.4683,
+      "step": 5197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.737803305351864e-06,
+      "loss": 2.4923,
+      "step": 5198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.734134260882511e-06,
+      "loss": 2.379,
+      "step": 5199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730465918295117e-06,
+      "loss": 2.4963,
+      "step": 5200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.726798278193241e-06,
+      "loss": 2.4173,
+      "step": 5201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7231313411803456e-06,
+      "loss": 2.4403,
+      "step": 5202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719465107859772e-06,
+      "loss": 2.4669,
+      "step": 5203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.715799578834731e-06,
+      "loss": 2.4251,
+      "step": 5204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.712134754708331e-06,
+      "loss": 2.4885,
+      "step": 5205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708470636083571e-06,
+      "loss": 2.5243,
+      "step": 5206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.704807223563315e-06,
+      "loss": 2.4305,
+      "step": 5207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701144517750323e-06,
+      "loss": 2.441,
+      "step": 5208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697482519247237e-06,
+      "loss": 2.3673,
+      "step": 5209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6938212286565835e-06,
+      "loss": 2.4476,
+      "step": 5210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.690160646580769e-06,
+      "loss": 2.4742,
+      "step": 5211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.686500773622089e-06,
+      "loss": 2.59,
+      "step": 5212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.682841610382709e-06,
+      "loss": 2.404,
+      "step": 5213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679183157464703e-06,
+      "loss": 2.5424,
+      "step": 5214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675525415469998e-06,
+      "loss": 2.4504,
+      "step": 5215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.671868385000425e-06,
+      "loss": 2.4329,
+      "step": 5216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.668212066657697e-06,
+      "loss": 2.4492,
+      "step": 5217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.664556461043389e-06,
+      "loss": 2.4713,
+      "step": 5218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.660901568758988e-06,
+      "loss": 2.427,
+      "step": 5219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.65724739040585e-06,
+      "loss": 2.459,
+      "step": 5220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653593926585204e-06,
+      "loss": 2.4857,
+      "step": 5221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.649941177898173e-06,
+      "loss": 2.4411,
+      "step": 5222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.646289144945769e-06,
+      "loss": 2.4316,
+      "step": 5223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642637828328868e-06,
+      "loss": 2.4822,
+      "step": 5224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638987228648238e-06,
+      "loss": 2.4739,
+      "step": 5225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6353373465045324e-06,
+      "loss": 2.4443,
+      "step": 5226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63168818249828e-06,
+      "loss": 2.4404,
+      "step": 5227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.628039737229896e-06,
+      "loss": 2.3806,
+      "step": 5228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62439201129968e-06,
+      "loss": 2.4523,
+      "step": 5229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.620745005307793e-06,
+      "loss": 2.4023,
+      "step": 5230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617098719854312e-06,
+      "loss": 2.4662,
+      "step": 5231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.613453155539164e-06,
+      "loss": 2.3819,
+      "step": 5232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.609808312962172e-06,
+      "loss": 2.572,
+      "step": 5233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606164192723043e-06,
+      "loss": 2.526,
+      "step": 5234
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602520795421357e-06,
+      "loss": 2.4374,
+      "step": 5235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.598878121656578e-06,
+      "loss": 2.5225,
+      "step": 5236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595236172028057e-06,
+      "loss": 2.4677,
+      "step": 5237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5915949471350115e-06,
+      "loss": 2.5523,
+      "step": 5238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587954447576549e-06,
+      "loss": 2.4583,
+      "step": 5239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584314673951669e-06,
+      "loss": 2.5361,
+      "step": 5240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.580675626859227e-06,
+      "loss": 2.4758,
+      "step": 5241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577037306897977e-06,
+      "loss": 2.4828,
+      "step": 5242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.573399714666545e-06,
+      "loss": 2.4066,
+      "step": 5243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5697628507634445e-06,
+      "loss": 2.4159,
+      "step": 5244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.566126715787063e-06,
+      "loss": 2.4395,
+      "step": 5245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562491310335675e-06,
+      "loss": 2.3689,
+      "step": 5246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5588566350074155e-06,
+      "loss": 2.4373,
+      "step": 5247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.555222690400333e-06,
+      "loss": 2.4898,
+      "step": 5248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.551589477112321e-06,
+      "loss": 2.4627,
+      "step": 5249
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547956995741176e-06,
+      "loss": 2.4564,
+      "step": 5250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.544325246884566e-06,
+      "loss": 2.4311,
+      "step": 5251
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.540694231140036e-06,
+      "loss": 2.4418,
+      "step": 5252
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.537063949105015e-06,
+      "loss": 2.5075,
+      "step": 5253
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.533434401376814e-06,
+      "loss": 2.4966,
+      "step": 5254
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529805588552609e-06,
+      "loss": 2.4511,
+      "step": 5255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.526177511229467e-06,
+      "loss": 2.4731,
+      "step": 5256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5225501700043425e-06,
+      "loss": 2.3644,
+      "step": 5257
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518923565474047e-06,
+      "loss": 2.5474,
+      "step": 5258
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.515297698235285e-06,
+      "loss": 2.3987,
+      "step": 5259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5116725688846375e-06,
+      "loss": 2.4597,
+      "step": 5260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.508048178018562e-06,
+      "loss": 2.4614,
+      "step": 5261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.504424526233397e-06,
+      "loss": 2.4612,
+      "step": 5262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500801614125362e-06,
+      "loss": 2.413,
+      "step": 5263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497179442290538e-06,
+      "loss": 2.3789,
+      "step": 5264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.493558011324913e-06,
+      "loss": 2.4346,
+      "step": 5265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489937321824328e-06,
+      "loss": 2.4913,
+      "step": 5266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4863173743845115e-06,
+      "loss": 2.3841,
+      "step": 5267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.482698169601072e-06,
+      "loss": 2.499,
+      "step": 5268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4790797080694926e-06,
+      "loss": 2.4451,
+      "step": 5269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.475461990385134e-06,
+      "loss": 2.6066,
+      "step": 5270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471845017143242e-06,
+      "loss": 2.4768,
+      "step": 5271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4682287889389185e-06,
+      "loss": 2.495,
+      "step": 5272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.464613306367172e-06,
+      "loss": 2.4679,
+      "step": 5273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.460998570022873e-06,
+      "loss": 2.4826,
+      "step": 5274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457384580500762e-06,
+      "loss": 2.4554,
+      "step": 5275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.453771338395468e-06,
+      "loss": 2.48,
+      "step": 5276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4501588443014964e-06,
+      "loss": 2.5417,
+      "step": 5277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.446547098813226e-06,
+      "loss": 2.4777,
+      "step": 5278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.442936102524911e-06,
+      "loss": 2.4718,
+      "step": 5279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.439325856030693e-06,
+      "loss": 2.4271,
+      "step": 5280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.435716359924568e-06,
+      "loss": 2.4304,
+      "step": 5281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432107614800437e-06,
+      "loss": 2.4228,
+      "step": 5282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4284996212520524e-06,
+      "loss": 2.4442,
+      "step": 5283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424892379873058e-06,
+      "loss": 2.4766,
+      "step": 5284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4212858912569685e-06,
+      "loss": 2.5569,
+      "step": 5285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.417680155997176e-06,
+      "loss": 2.497,
+      "step": 5286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414075174686948e-06,
+      "loss": 2.35,
+      "step": 5287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410470947919432e-06,
+      "loss": 2.5486,
+      "step": 5288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406867476287636e-06,
+      "loss": 2.4631,
+      "step": 5289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.403264760384465e-06,
+      "loss": 2.4172,
+      "step": 5290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399662800802695e-06,
+      "loss": 2.5102,
+      "step": 5291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396061598134961e-06,
+      "loss": 2.4887,
+      "step": 5292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.392461152973789e-06,
+      "loss": 2.5049,
+      "step": 5293
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3888614659115765e-06,
+      "loss": 2.5678,
+      "step": 5294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3852625375405975e-06,
+      "loss": 2.5588,
+      "step": 5295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.381664368452998e-06,
+      "loss": 2.4808,
+      "step": 5296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378066959240807e-06,
+      "loss": 2.4951,
+      "step": 5297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3744703104959065e-06,
+      "loss": 2.4776,
+      "step": 5298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.370874422810088e-06,
+      "loss": 2.4124,
+      "step": 5299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367279296774987e-06,
+      "loss": 2.4283,
+      "step": 5300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36368493298213e-06,
+      "loss": 2.4923,
+      "step": 5301
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360091332022912e-06,
+      "loss": 2.4491,
+      "step": 5302
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356498494488607e-06,
+      "loss": 2.5224,
+      "step": 5303
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352906420970356e-06,
+      "loss": 2.499,
+      "step": 5304
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349315112059188e-06,
+      "loss": 2.5186,
+      "step": 5305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345724568345981e-06,
+      "loss": 2.5383,
+      "step": 5306
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.342134790421519e-06,
+      "loss": 2.4944,
+      "step": 5307
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338545778876441e-06,
+      "loss": 2.4212,
+      "step": 5308
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.334957534301255e-06,
+      "loss": 2.4086,
+      "step": 5309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.331370057286358e-06,
+      "loss": 2.5035,
+      "step": 5310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.327783348422011e-06,
+      "loss": 2.3885,
+      "step": 5311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324197408298353e-06,
+      "loss": 2.4676,
+      "step": 5312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.320612237505394e-06,
+      "loss": 2.5116,
+      "step": 5313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.317027836633021e-06,
+      "loss": 2.4843,
+      "step": 5314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.313444206270981e-06,
+      "loss": 2.5226,
+      "step": 5315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.30986134700892e-06,
+      "loss": 2.5199,
+      "step": 5316
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.306279259436331e-06,
+      "loss": 2.4124,
+      "step": 5317
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302697944142593e-06,
+      "loss": 2.3847,
+      "step": 5318
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299117401716958e-06,
+      "loss": 2.4657,
+      "step": 5319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.295537632748546e-06,
+      "loss": 2.5261,
+      "step": 5320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291958637826355e-06,
+      "loss": 2.496,
+      "step": 5321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288380417539255e-06,
+      "loss": 2.521,
+      "step": 5322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284802972475975e-06,
+      "loss": 2.6107,
+      "step": 5323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281226303225142e-06,
+      "loss": 2.3171,
+      "step": 5324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277650410375239e-06,
+      "loss": 2.4015,
+      "step": 5325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274075294514617e-06,
+      "loss": 2.5191,
+      "step": 5326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.270500956231508e-06,
+      "loss": 2.3762,
+      "step": 5327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.266927396114017e-06,
+      "loss": 2.466,
+      "step": 5328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263354614750115e-06,
+      "loss": 2.4307,
+      "step": 5329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.259782612727649e-06,
+      "loss": 2.4653,
+      "step": 5330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256211390634335e-06,
+      "loss": 2.4951,
+      "step": 5331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252640949057763e-06,
+      "loss": 2.4855,
+      "step": 5332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249071288585399e-06,
+      "loss": 2.4994,
+      "step": 5333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.245502409804566e-06,
+      "loss": 2.4578,
+      "step": 5334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.241934313302472e-06,
+      "loss": 2.507,
+      "step": 5335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238366999666191e-06,
+      "loss": 2.5154,
+      "step": 5336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23480046948267e-06,
+      "loss": 2.4903,
+      "step": 5337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231234723338725e-06,
+      "loss": 2.4655,
+      "step": 5338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22766976182105e-06,
+      "loss": 2.4245,
+      "step": 5339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22410558551619e-06,
+      "loss": 2.4648,
+      "step": 5340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.220542195010588e-06,
+      "loss": 2.499,
+      "step": 5341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.216979590890545e-06,
+      "loss": 2.5071,
+      "step": 5342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.213417773742224e-06,
+      "loss": 2.4987,
+      "step": 5343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209856744151664e-06,
+      "loss": 2.4236,
+      "step": 5344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206296502704794e-06,
+      "loss": 2.3688,
+      "step": 5345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202737049987381e-06,
+      "loss": 2.449,
+      "step": 5346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.199178386585082e-06,
+      "loss": 2.5006,
+      "step": 5347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.19562051308342e-06,
+      "loss": 2.5004,
+      "step": 5348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192063430067788e-06,
+      "loss": 2.4056,
+      "step": 5349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188507138123456e-06,
+      "loss": 2.4162,
+      "step": 5350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184951637835541e-06,
+      "loss": 2.4358,
+      "step": 5351
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.181396929789056e-06,
+      "loss": 2.527,
+      "step": 5352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.177843014568871e-06,
+      "loss": 2.4253,
+      "step": 5353
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1742898927597275e-06,
+      "loss": 2.4841,
+      "step": 5354
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1707375649462355e-06,
+      "loss": 2.4891,
+      "step": 5355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.167186031712881e-06,
+      "loss": 2.5443,
+      "step": 5356
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163635293644e-06,
+      "loss": 2.4622,
+      "step": 5357
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.160085351323825e-06,
+      "loss": 2.3781,
+      "step": 5358
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.156536205336443e-06,
+      "loss": 2.508,
+      "step": 5359
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152987856265803e-06,
+      "loss": 2.4991,
+      "step": 5360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1494403046957316e-06,
+      "loss": 2.3646,
+      "step": 5361
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.145893551209933e-06,
+      "loss": 2.5949,
+      "step": 5362
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14234759639196e-06,
+      "loss": 2.5252,
+      "step": 5363
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138802440825248e-06,
+      "loss": 2.3998,
+      "step": 5364
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135258085093099e-06,
+      "loss": 2.4189,
+      "step": 5365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.131714529778678e-06,
+      "loss": 2.5266,
+      "step": 5366
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12817177546503e-06,
+      "loss": 2.4707,
+      "step": 5367
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12462982273505e-06,
+      "loss": 2.452,
+      "step": 5368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12108867217151e-06,
+      "loss": 2.3943,
+      "step": 5369
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117548324357067e-06,
+      "loss": 2.4605,
+      "step": 5370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.114008779874215e-06,
+      "loss": 2.4143,
+      "step": 5371
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1104700393053375e-06,
+      "loss": 2.5031,
+      "step": 5372
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.106932103232683e-06,
+      "loss": 2.469,
+      "step": 5373
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103394972238349e-06,
+      "loss": 2.4568,
+      "step": 5374
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.09985864690433e-06,
+      "loss": 2.4788,
+      "step": 5375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.096323127812473e-06,
+      "loss": 2.4028,
+      "step": 5376
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092788415544485e-06,
+      "loss": 2.5105,
+      "step": 5377
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089254510681948e-06,
+      "loss": 2.4621,
+      "step": 5378
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.085721413806322e-06,
+      "loss": 2.4331,
+      "step": 5379
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.082189125498913e-06,
+      "loss": 2.4848,
+      "step": 5380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078657646340908e-06,
+      "loss": 2.4295,
+      "step": 5381
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0751269769133534e-06,
+      "loss": 2.6119,
+      "step": 5382
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.071597117797171e-06,
+      "loss": 2.4513,
+      "step": 5383
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.068068069573142e-06,
+      "loss": 2.4397,
+      "step": 5384
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064539832821922e-06,
+      "loss": 2.5242,
+      "step": 5385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.061012408124013e-06,
+      "loss": 2.5581,
+      "step": 5386
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.057485796059816e-06,
+      "loss": 2.5142,
+      "step": 5387
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.053959997209565e-06,
+      "loss": 2.4858,
+      "step": 5388
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.050435012153385e-06,
+      "loss": 2.3541,
+      "step": 5389
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.046910841471253e-06,
+      "loss": 2.4421,
+      "step": 5390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043387485743019e-06,
+      "loss": 2.4057,
+      "step": 5391
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.039864945548394e-06,
+      "loss": 2.4319,
+      "step": 5392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.036343221466964e-06,
+      "loss": 2.5008,
+      "step": 5393
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032822314078165e-06,
+      "loss": 2.4599,
+      "step": 5394
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0293022239613066e-06,
+      "loss": 2.4008,
+      "step": 5395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.025782951695577e-06,
+      "loss": 2.4942,
+      "step": 5396
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.022264497860007e-06,
+      "loss": 2.4882,
+      "step": 5397
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018746863033509e-06,
+      "loss": 2.4736,
+      "step": 5398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0152300477948524e-06,
+      "loss": 2.5393,
+      "step": 5399
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0117140527226745e-06,
+      "loss": 2.4392,
+      "step": 5400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.008198878395479e-06,
+      "loss": 2.4592,
+      "step": 5401
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.004684525391637e-06,
+      "loss": 2.406,
+      "step": 5402
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.001170994289369e-06,
+      "loss": 2.4362,
+      "step": 5403
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.997658285666787e-06,
+      "loss": 2.37,
+      "step": 5404
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994146400101841e-06,
+      "loss": 2.4914,
+      "step": 5405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.990635338172361e-06,
+      "loss": 2.448,
+      "step": 5406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987125100456039e-06,
+      "loss": 2.5009,
+      "step": 5407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.98361568753043e-06,
+      "loss": 2.4972,
+      "step": 5408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980107099972951e-06,
+      "loss": 2.431,
+      "step": 5409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976599338360893e-06,
+      "loss": 2.5143,
+      "step": 5410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973092403271391e-06,
+      "loss": 2.4901,
+      "step": 5411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969586295281462e-06,
+      "loss": 2.4375,
+      "step": 5412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966081014967991e-06,
+      "loss": 2.5361,
+      "step": 5413
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962576562907705e-06,
+      "loss": 2.5361,
+      "step": 5414
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.959072939677212e-06,
+      "loss": 2.4316,
+      "step": 5415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.955570145852978e-06,
+      "loss": 2.4876,
+      "step": 5416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9520681820113345e-06,
+      "loss": 2.3787,
+      "step": 5417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948567048728476e-06,
+      "loss": 2.4614,
+      "step": 5418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94506674658046e-06,
+      "loss": 2.465,
+      "step": 5419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.941567276143199e-06,
+      "loss": 2.4931,
+      "step": 5420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9380686379924895e-06,
+      "loss": 2.4342,
+      "step": 5421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.934570832703968e-06,
+      "loss": 2.467,
+      "step": 5422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931073860853146e-06,
+      "loss": 2.4857,
+      "step": 5423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.927577723015399e-06,
+      "loss": 2.4437,
+      "step": 5424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924082419765959e-06,
+      "loss": 2.2986,
+      "step": 5425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.920587951679926e-06,
+      "loss": 2.4245,
+      "step": 5426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917094319332263e-06,
+      "loss": 2.4995,
+      "step": 5427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913601523297783e-06,
+      "loss": 2.4141,
+      "step": 5428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.910109564151181e-06,
+      "loss": 2.5067,
+      "step": 5429
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.906618442467005e-06,
+      "loss": 2.4838,
+      "step": 5430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903128158819659e-06,
+      "loss": 2.3731,
+      "step": 5431
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8996387137834155e-06,
+      "loss": 2.4892,
+      "step": 5432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.896150107932412e-06,
+      "loss": 2.4763,
+      "step": 5433
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892662341840642e-06,
+      "loss": 2.4732,
+      "step": 5434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889175416081965e-06,
+      "loss": 2.4802,
+      "step": 5435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885689331230103e-06,
+      "loss": 2.394,
+      "step": 5436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.882204087858626e-06,
+      "loss": 2.474,
+      "step": 5437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.878719686540992e-06,
+      "loss": 2.5448,
+      "step": 5438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875236127850494e-06,
+      "loss": 2.5006,
+      "step": 5439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8717534123603e-06,
+      "loss": 2.4943,
+      "step": 5440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8682715406434386e-06,
+      "loss": 2.5107,
+      "step": 5441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8647905132727965e-06,
+      "loss": 2.4601,
+      "step": 5442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8613103308211226e-06,
+      "loss": 2.4657,
+      "step": 5443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857830993861032e-06,
+      "loss": 2.4734,
+      "step": 5444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854352502964983e-06,
+      "loss": 2.4929,
+      "step": 5445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.850874858705319e-06,
+      "loss": 2.436,
+      "step": 5446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847398061654231e-06,
+      "loss": 2.4234,
+      "step": 5447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.843922112383768e-06,
+      "loss": 2.4985,
+      "step": 5448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840447011465843e-06,
+      "loss": 2.4457,
+      "step": 5449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836972759472232e-06,
+      "loss": 2.4855,
+      "step": 5450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.83349935697457e-06,
+      "loss": 2.5098,
+      "step": 5451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.83002680454435e-06,
+      "loss": 2.4292,
+      "step": 5452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.82655510275293e-06,
+      "loss": 2.4534,
+      "step": 5453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8230842521715145e-06,
+      "loss": 2.5029,
+      "step": 5454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.819614253371193e-06,
+      "loss": 2.4193,
+      "step": 5455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8161451069228884e-06,
+      "loss": 2.4785,
+      "step": 5456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8126768133973975e-06,
+      "loss": 2.3951,
+      "step": 5457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.809209373365377e-06,
+      "loss": 2.3903,
+      "step": 5458
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.805742787397337e-06,
+      "loss": 2.4432,
+      "step": 5459
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.802277056063652e-06,
+      "loss": 2.4968,
+      "step": 5460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7988121799345585e-06,
+      "loss": 2.4568,
+      "step": 5461
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.795348159580136e-06,
+      "loss": 2.354,
+      "step": 5462
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.791884995570346e-06,
+      "loss": 2.4908,
+      "step": 5463
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788422688475e-06,
+      "loss": 2.519,
+      "step": 5464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.784961238863758e-06,
+      "loss": 2.4469,
+      "step": 5465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.781500647306152e-06,
+      "loss": 2.4914,
+      "step": 5466
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.778040914371569e-06,
+      "loss": 2.3438,
+      "step": 5467
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774582040629253e-06,
+      "loss": 2.4578,
+      "step": 5468
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771124026648309e-06,
+      "loss": 2.4347,
+      "step": 5469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.767666872997704e-06,
+      "loss": 2.5331,
+      "step": 5470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764210580246246e-06,
+      "loss": 2.4278,
+      "step": 5471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.760755148962629e-06,
+      "loss": 2.5758,
+      "step": 5472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757300579715382e-06,
+      "loss": 2.4331,
+      "step": 5473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.753846873072902e-06,
+      "loss": 2.4905,
+      "step": 5474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.750394029603444e-06,
+      "loss": 2.4727,
+      "step": 5475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.746942049875119e-06,
+      "loss": 2.4254,
+      "step": 5476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.743490934455897e-06,
+      "loss": 2.4863,
+      "step": 5477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7400406839136095e-06,
+      "loss": 2.4353,
+      "step": 5478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.73659129881593e-06,
+      "loss": 2.4873,
+      "step": 5479
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.733142779730413e-06,
+      "loss": 2.3903,
+      "step": 5480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.729695127224457e-06,
+      "loss": 2.5284,
+      "step": 5481
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726248341865315e-06,
+      "loss": 2.5125,
+      "step": 5482
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722802424220104e-06,
+      "loss": 2.465,
+      "step": 5483
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719357374855796e-06,
+      "loss": 2.473,
+      "step": 5484
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7159131943392225e-06,
+      "loss": 2.438,
+      "step": 5485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.712469883237066e-06,
+      "loss": 2.4183,
+      "step": 5486
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.709027442115874e-06,
+      "loss": 2.458,
+      "step": 5487
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705585871542045e-06,
+      "loss": 2.4442,
+      "step": 5488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.702145172081838e-06,
+      "loss": 2.5816,
+      "step": 5489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.698705344301362e-06,
+      "loss": 2.4454,
+      "step": 5490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695266388766589e-06,
+      "loss": 2.5563,
+      "step": 5491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691828306043347e-06,
+      "loss": 2.5116,
+      "step": 5492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.688391096697319e-06,
+      "loss": 2.4575,
+      "step": 5493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684954761294043e-06,
+      "loss": 2.3991,
+      "step": 5494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6815193003989204e-06,
+      "loss": 2.4529,
+      "step": 5495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.67808471457719e-06,
+      "loss": 2.5169,
+      "step": 5496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674651004393972e-06,
+      "loss": 2.434,
+      "step": 5497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671218170414229e-06,
+      "loss": 2.51,
+      "step": 5498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667786213202774e-06,
+      "loss": 2.5077,
+      "step": 5499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.664355133324286e-06,
+      "loss": 2.4859,
+      "step": 5500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660924931343296e-06,
+      "loss": 2.51,
+      "step": 5501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657495607824188e-06,
+      "loss": 2.5241,
+      "step": 5502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654067163331207e-06,
+      "loss": 2.4509,
+      "step": 5503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65063959842845e-06,
+      "loss": 2.5026,
+      "step": 5504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647212913679868e-06,
+      "loss": 2.5035,
+      "step": 5505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.643787109649274e-06,
+      "loss": 2.421,
+      "step": 5506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.640362186900321e-06,
+      "loss": 2.4681,
+      "step": 5507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.636938145996534e-06,
+      "loss": 2.3666,
+      "step": 5508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6335149875012834e-06,
+      "loss": 2.4705,
+      "step": 5509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630092711977797e-06,
+      "loss": 2.3666,
+      "step": 5510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626671319989157e-06,
+      "loss": 2.554,
+      "step": 5511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.623250812098305e-06,
+      "loss": 2.4935,
+      "step": 5512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619831188868021e-06,
+      "loss": 2.4532,
+      "step": 5513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.616412450860963e-06,
+      "loss": 2.4652,
+      "step": 5514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.612994598639628e-06,
+      "loss": 2.4611,
+      "step": 5515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.609577632766367e-06,
+      "loss": 2.4258,
+      "step": 5516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6061615538033885e-06,
+      "loss": 2.4412,
+      "step": 5517
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.602746362312759e-06,
+      "loss": 2.4658,
+      "step": 5518
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599332058856393e-06,
+      "loss": 2.3873,
+      "step": 5519
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595918643996061e-06,
+      "loss": 2.502,
+      "step": 5520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5925061182933885e-06,
+      "loss": 2.4203,
+      "step": 5521
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.589094482309853e-06,
+      "loss": 2.5636,
+      "step": 5522
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585683736606789e-06,
+      "loss": 2.5161,
+      "step": 5523
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582273881745377e-06,
+      "loss": 2.4597,
+      "step": 5524
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.578864918286654e-06,
+      "loss": 2.5072,
+      "step": 5525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.575456846791523e-06,
+      "loss": 2.4679,
+      "step": 5526
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.572049667820717e-06,
+      "loss": 2.4996,
+      "step": 5527
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.56864338193484e-06,
+      "loss": 2.4981,
+      "step": 5528
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565237989694346e-06,
+      "loss": 2.4683,
+      "step": 5529
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.56183349165953e-06,
+      "loss": 2.5645,
+      "step": 5530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558429888390559e-06,
+      "loss": 2.4579,
+      "step": 5531
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.555027180447442e-06,
+      "loss": 2.4025,
+      "step": 5532
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551625368390036e-06,
+      "loss": 2.4766,
+      "step": 5533
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.548224452778056e-06,
+      "loss": 2.5322,
+      "step": 5534
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.544824434171082e-06,
+      "loss": 2.4779,
+      "step": 5535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.54142531312852e-06,
+      "loss": 2.4898,
+      "step": 5536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538027090209649e-06,
+      "loss": 2.4541,
+      "step": 5537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.534629765973594e-06,
+      "loss": 2.4097,
+      "step": 5538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5312333409793295e-06,
+      "loss": 2.5778,
+      "step": 5539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5278378157856905e-06,
+      "loss": 2.4903,
+      "step": 5540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5244431909513506e-06,
+      "loss": 2.4096,
+      "step": 5541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5210494670348406e-06,
+      "loss": 2.4824,
+      "step": 5542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.517656644594558e-06,
+      "loss": 2.3986,
+      "step": 5543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.514264724188727e-06,
+      "loss": 2.4478,
+      "step": 5544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.510873706375439e-06,
+      "loss": 2.5093,
+      "step": 5545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507483591712633e-06,
+      "loss": 2.4868,
+      "step": 5546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504094380758103e-06,
+      "loss": 2.3903,
+      "step": 5547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500706074069486e-06,
+      "loss": 2.4806,
+      "step": 5548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497318672204282e-06,
+      "loss": 2.4459,
+      "step": 5549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.493932175719827e-06,
+      "loss": 2.5235,
+      "step": 5550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.490546585173318e-06,
+      "loss": 2.4083,
+      "step": 5551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.487161901121811e-06,
+      "loss": 2.4297,
+      "step": 5552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.483778124122191e-06,
+      "loss": 2.444,
+      "step": 5553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4803952547312125e-06,
+      "loss": 2.4594,
+      "step": 5554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4770132935054715e-06,
+      "loss": 2.497,
+      "step": 5555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.473632241001417e-06,
+      "loss": 2.4564,
+      "step": 5556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470252097775356e-06,
+      "loss": 2.4851,
+      "step": 5557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466872864383428e-06,
+      "loss": 2.424,
+      "step": 5558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463494541381634e-06,
+      "loss": 2.4668,
+      "step": 5559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.460117129325836e-06,
+      "loss": 2.494,
+      "step": 5560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.456740628771724e-06,
+      "loss": 2.449,
+      "step": 5561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.453365040274852e-06,
+      "loss": 2.5295,
+      "step": 5562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449990364390619e-06,
+      "loss": 2.49,
+      "step": 5563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.446616601674278e-06,
+      "loss": 2.5519,
+      "step": 5564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443243752680929e-06,
+      "loss": 2.448,
+      "step": 5565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439871817965525e-06,
+      "loss": 2.4862,
+      "step": 5566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4365007980828554e-06,
+      "loss": 2.5043,
+      "step": 5567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.433130693587574e-06,
+      "loss": 2.5177,
+      "step": 5568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.429761505034188e-06,
+      "loss": 2.4319,
+      "step": 5569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426393232977034e-06,
+      "loss": 2.4734,
+      "step": 5570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423025877970311e-06,
+      "loss": 2.4338,
+      "step": 5571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.419659440568069e-06,
+      "loss": 2.5198,
+      "step": 5572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416293921324199e-06,
+      "loss": 2.4065,
+      "step": 5573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.41292932079245e-06,
+      "loss": 2.4777,
+      "step": 5574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.409565639526409e-06,
+      "loss": 2.4029,
+      "step": 5575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406202878079515e-06,
+      "loss": 2.5393,
+      "step": 5576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4028410370050705e-06,
+      "loss": 2.4225,
+      "step": 5577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399480116856203e-06,
+      "loss": 2.5417,
+      "step": 5578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.396120118185902e-06,
+      "loss": 2.517,
+      "step": 5579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392761041547006e-06,
+      "loss": 2.3999,
+      "step": 5580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.389402887492197e-06,
+      "loss": 2.4256,
+      "step": 5581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386045656574008e-06,
+      "loss": 2.5,
+      "step": 5582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382689349344821e-06,
+      "loss": 2.4874,
+      "step": 5583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379333966356855e-06,
+      "loss": 2.4683,
+      "step": 5584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.375979508162197e-06,
+      "loss": 2.5277,
+      "step": 5585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3726259753127705e-06,
+      "loss": 2.4945,
+      "step": 5586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36927336836034e-06,
+      "loss": 2.4631,
+      "step": 5587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3659216878565295e-06,
+      "loss": 2.428,
+      "step": 5588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.362570934352805e-06,
+      "loss": 2.4959,
+      "step": 5589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35922110840048e-06,
+      "loss": 2.4482,
+      "step": 5590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355872210550718e-06,
+      "loss": 2.4823,
+      "step": 5591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.352524241354531e-06,
+      "loss": 2.4534,
+      "step": 5592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349177201362765e-06,
+      "loss": 2.4332,
+      "step": 5593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.345831091126138e-06,
+      "loss": 2.3755,
+      "step": 5594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342485911195189e-06,
+      "loss": 2.3636,
+      "step": 5595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.339141662120319e-06,
+      "loss": 2.4965,
+      "step": 5596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335798344451773e-06,
+      "loss": 2.4947,
+      "step": 5597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3324559587396395e-06,
+      "loss": 2.4635,
+      "step": 5598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3291145055338605e-06,
+      "loss": 2.4507,
+      "step": 5599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.325773985384221e-06,
+      "loss": 2.375,
+      "step": 5600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3224343988403405e-06,
+      "loss": 2.4429,
+      "step": 5601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319095746451708e-06,
+      "loss": 2.4496,
+      "step": 5602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315758028767647e-06,
+      "loss": 2.4901,
+      "step": 5603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31242124633732e-06,
+      "loss": 2.4323,
+      "step": 5604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309085399709745e-06,
+      "loss": 2.5063,
+      "step": 5605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305750489433786e-06,
+      "loss": 2.4973,
+      "step": 5606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302416516058148e-06,
+      "loss": 2.4837,
+      "step": 5607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.299083480131386e-06,
+      "loss": 2.4527,
+      "step": 5608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2957513822019035e-06,
+      "loss": 2.5096,
+      "step": 5609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292420222817934e-06,
+      "loss": 2.4432,
+      "step": 5610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.289090002527581e-06,
+      "loss": 2.4904,
+      "step": 5611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.285760721878774e-06,
+      "loss": 2.5396,
+      "step": 5612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282432381419293e-06,
+      "loss": 2.4873,
+      "step": 5613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279104981696766e-06,
+      "loss": 2.4387,
+      "step": 5614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275778523258668e-06,
+      "loss": 2.4695,
+      "step": 5615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272453006652313e-06,
+      "loss": 2.439,
+      "step": 5616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2691284324248685e-06,
+      "loss": 2.4775,
+      "step": 5617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.265804801123329e-06,
+      "loss": 2.4806,
+      "step": 5618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262482113294558e-06,
+      "loss": 2.478,
+      "step": 5619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259160369485252e-06,
+      "loss": 2.5105,
+      "step": 5620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.255839570241947e-06,
+      "loss": 2.3818,
+      "step": 5621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2525197161110285e-06,
+      "loss": 2.4417,
+      "step": 5622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2492008076387314e-06,
+      "loss": 2.4289,
+      "step": 5623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.245882845371127e-06,
+      "loss": 2.4337,
+      "step": 5624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242565829854138e-06,
+      "loss": 2.4955,
+      "step": 5625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2392497616335285e-06,
+      "loss": 2.4814,
+      "step": 5626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.235934641254896e-06,
+      "loss": 2.4858,
+      "step": 5627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232620469263707e-06,
+      "loss": 2.501,
+      "step": 5628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.229307246205245e-06,
+      "loss": 2.4156,
+      "step": 5629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225994972624655e-06,
+      "loss": 2.4702,
+      "step": 5630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222683649066918e-06,
+      "loss": 2.4231,
+      "step": 5631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.219373276076862e-06,
+      "loss": 2.5667,
+      "step": 5632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.216063854199158e-06,
+      "loss": 2.5233,
+      "step": 5633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.212755383978323e-06,
+      "loss": 2.5014,
+      "step": 5634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2094478659587026e-06,
+      "loss": 2.49,
+      "step": 5635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.20614130068451e-06,
+      "loss": 2.5383,
+      "step": 5636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.202835688699788e-06,
+      "loss": 2.4595,
+      "step": 5637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.199531030548417e-06,
+      "loss": 2.4811,
+      "step": 5638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1962273267741305e-06,
+      "loss": 2.4552,
+      "step": 5639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192924577920502e-06,
+      "loss": 2.4434,
+      "step": 5640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1896227845309475e-06,
+      "loss": 2.5524,
+      "step": 5641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.186321947148726e-06,
+      "loss": 2.4887,
+      "step": 5642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1830220663169375e-06,
+      "loss": 2.4438,
+      "step": 5643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179723142578528e-06,
+      "loss": 2.4202,
+      "step": 5644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176425176476288e-06,
+      "loss": 2.4992,
+      "step": 5645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.173128168552838e-06,
+      "loss": 2.4642,
+      "step": 5646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169832119350653e-06,
+      "loss": 2.4443,
+      "step": 5647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166537029412048e-06,
+      "loss": 2.4697,
+      "step": 5648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163242899279177e-06,
+      "loss": 2.4607,
+      "step": 5649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1599497294940405e-06,
+      "loss": 2.422,
+      "step": 5650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156657520598479e-06,
+      "loss": 2.4382,
+      "step": 5651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.153366273134167e-06,
+      "loss": 2.5234,
+      "step": 5652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150075987642637e-06,
+      "loss": 2.5506,
+      "step": 5653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1467866646652535e-06,
+      "loss": 2.4608,
+      "step": 5654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.143498304743217e-06,
+      "loss": 2.4032,
+      "step": 5655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.140210908417581e-06,
+      "loss": 2.5141,
+      "step": 5656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.136924476229235e-06,
+      "loss": 2.4676,
+      "step": 5657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.13363900871891e-06,
+      "loss": 2.4588,
+      "step": 5658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130354506427177e-06,
+      "loss": 2.4574,
+      "step": 5659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.127070969894452e-06,
+      "loss": 2.4636,
+      "step": 5660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12378839966099e-06,
+      "loss": 2.471,
+      "step": 5661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1205067962668885e-06,
+      "loss": 2.4023,
+      "step": 5662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1172261602520806e-06,
+      "loss": 2.4743,
+      "step": 5663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.113946492156343e-06,
+      "loss": 2.4281,
+      "step": 5664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1106677925192985e-06,
+      "loss": 2.4616,
+      "step": 5665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1073900618804045e-06,
+      "loss": 2.461,
+      "step": 5666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.10411330077896e-06,
+      "loss": 2.5039,
+      "step": 5667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1008375097541086e-06,
+      "loss": 2.416,
+      "step": 5668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.097562689344822e-06,
+      "loss": 2.4187,
+      "step": 5669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0942888400899304e-06,
+      "loss": 2.5282,
+      "step": 5670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091015962528094e-06,
+      "loss": 2.4,
+      "step": 5671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.08774405719781e-06,
+      "loss": 2.4267,
+      "step": 5672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.084473124637419e-06,
+      "loss": 2.4068,
+      "step": 5673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.081203165385105e-06,
+      "loss": 2.4835,
+      "step": 5674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.077934179978889e-06,
+      "loss": 2.5449,
+      "step": 5675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074666168956629e-06,
+      "loss": 2.4995,
+      "step": 5676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.071399132856029e-06,
+      "loss": 2.3979,
+      "step": 5677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068133072214627e-06,
+      "loss": 2.4221,
+      "step": 5678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064867987569807e-06,
+      "loss": 2.4527,
+      "step": 5679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.06160387945878e-06,
+      "loss": 2.486,
+      "step": 5680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.058340748418604e-06,
+      "loss": 2.4658,
+      "step": 5681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.055078594986189e-06,
+      "loss": 2.5254,
+      "step": 5682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051817419698261e-06,
+      "loss": 2.3891,
+      "step": 5683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048557223091396e-06,
+      "loss": 2.4877,
+      "step": 5684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.045298005702016e-06,
+      "loss": 2.5544,
+      "step": 5685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042039768066362e-06,
+      "loss": 2.4347,
+      "step": 5686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038782510720538e-06,
+      "loss": 2.4197,
+      "step": 5687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.035526234200474e-06,
+      "loss": 2.3474,
+      "step": 5688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.032270939041935e-06,
+      "loss": 2.5736,
+      "step": 5689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029016625780529e-06,
+      "loss": 2.4311,
+      "step": 5690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0257632949517056e-06,
+      "loss": 2.4812,
+      "step": 5691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0225109470907494e-06,
+      "loss": 2.4147,
+      "step": 5692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0192595827327825e-06,
+      "loss": 2.4555,
+      "step": 5693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.016009202412769e-06,
+      "loss": 2.5068,
+      "step": 5694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0127598066655045e-06,
+      "loss": 2.4034,
+      "step": 5695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009511396025635e-06,
+      "loss": 2.401,
+      "step": 5696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.006263971027624e-06,
+      "loss": 2.4925,
+      "step": 5697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0030175322057885e-06,
+      "loss": 2.428,
+      "step": 5698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99977208009429e-06,
+      "loss": 2.4795,
+      "step": 5699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.996527615227104e-06,
+      "loss": 2.5042,
+      "step": 5700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9932841381380615e-06,
+      "loss": 2.514,
+      "step": 5701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990041649360828e-06,
+      "loss": 2.4428,
+      "step": 5702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986800149428903e-06,
+      "loss": 2.4244,
+      "step": 5703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983559638875626e-06,
+      "loss": 2.3991,
+      "step": 5704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.980320118234174e-06,
+      "loss": 2.4839,
+      "step": 5705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977081588037556e-06,
+      "loss": 2.5064,
+      "step": 5706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973844048818622e-06,
+      "loss": 2.4554,
+      "step": 5707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970607501110062e-06,
+      "loss": 2.4495,
+      "step": 5708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.967371945444397e-06,
+      "loss": 2.4485,
+      "step": 5709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.96413738235399e-06,
+      "loss": 2.4342,
+      "step": 5710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9609038123710355e-06,
+      "loss": 2.489,
+      "step": 5711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957671236027569e-06,
+      "loss": 2.4854,
+      "step": 5712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954439653855463e-06,
+      "loss": 2.5162,
+      "step": 5713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.951209066386419e-06,
+      "loss": 2.3945,
+      "step": 5714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.947979474151977e-06,
+      "loss": 2.3293,
+      "step": 5715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9447508776835305e-06,
+      "loss": 2.499,
+      "step": 5716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.941523277512281e-06,
+      "loss": 2.4669,
+      "step": 5717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.938296674169286e-06,
+      "loss": 2.4541,
+      "step": 5718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935071068185432e-06,
+      "loss": 2.5016,
+      "step": 5719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931846460091442e-06,
+      "loss": 2.4346,
+      "step": 5720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.928622850417875e-06,
+      "loss": 2.4713,
+      "step": 5721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.92540023969513e-06,
+      "loss": 2.3889,
+      "step": 5722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922178628453431e-06,
+      "loss": 2.498,
+      "step": 5723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.918958017222847e-06,
+      "loss": 2.4037,
+      "step": 5724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.915738406533278e-06,
+      "loss": 2.4544,
+      "step": 5725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.912519796914465e-06,
+      "loss": 2.4958,
+      "step": 5726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909302188895976e-06,
+      "loss": 2.5169,
+      "step": 5727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.906085583007221e-06,
+      "loss": 2.4358,
+      "step": 5728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9028699797774416e-06,
+      "loss": 2.4329,
+      "step": 5729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8996553797357186e-06,
+      "loss": 2.4683,
+      "step": 5730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.896441783410958e-06,
+      "loss": 2.5235,
+      "step": 5731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893229191331907e-06,
+      "loss": 2.4597,
+      "step": 5732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890017604027158e-06,
+      "loss": 2.4814,
+      "step": 5733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886807022025119e-06,
+      "loss": 2.3685,
+      "step": 5734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8835974458540425e-06,
+      "loss": 2.5189,
+      "step": 5735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8803888760420146e-06,
+      "loss": 2.5159,
+      "step": 5736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877181313116958e-06,
+      "loss": 2.4996,
+      "step": 5737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873974757606626e-06,
+      "loss": 2.4717,
+      "step": 5738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.87076921003861e-06,
+      "loss": 2.5099,
+      "step": 5739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8675646709403235e-06,
+      "loss": 2.5383,
+      "step": 5740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8643611408390334e-06,
+      "loss": 2.498,
+      "step": 5741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.861158620261832e-06,
+      "loss": 2.4292,
+      "step": 5742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8579571097356374e-06,
+      "loss": 2.4169,
+      "step": 5743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854756609787212e-06,
+      "loss": 2.5433,
+      "step": 5744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851557120943146e-06,
+      "loss": 2.5088,
+      "step": 5745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848358643729868e-06,
+      "loss": 2.5206,
+      "step": 5746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.84516117867364e-06,
+      "loss": 2.4811,
+      "step": 5747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.84196472630055e-06,
+      "loss": 2.4604,
+      "step": 5748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.838769287136523e-06,
+      "loss": 2.4337,
+      "step": 5749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.835574861707329e-06,
+      "loss": 2.4948,
+      "step": 5750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832381450538551e-06,
+      "loss": 2.5054,
+      "step": 5751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8291890541556194e-06,
+      "loss": 2.5244,
+      "step": 5752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825997673083794e-06,
+      "loss": 2.4112,
+      "step": 5753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822807307848165e-06,
+      "loss": 2.4952,
+      "step": 5754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819617958973659e-06,
+      "loss": 2.464,
+      "step": 5755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816429626985038e-06,
+      "loss": 2.4934,
+      "step": 5756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8132423124068794e-06,
+      "loss": 2.5437,
+      "step": 5757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8100560157636192e-06,
+      "loss": 2.4441,
+      "step": 5758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806870737579511e-06,
+      "loss": 2.401,
+      "step": 5759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8036864783786377e-06,
+      "loss": 2.4643,
+      "step": 5760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8005032386849227e-06,
+      "loss": 2.4425,
+      "step": 5761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797321019022118e-06,
+      "loss": 2.4626,
+      "step": 5762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794139819913809e-06,
+      "loss": 2.39,
+      "step": 5763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7909596418834162e-06,
+      "loss": 2.5297,
+      "step": 5764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.787780485454181e-06,
+      "loss": 2.4969,
+      "step": 5765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.784602351149185e-06,
+      "loss": 2.5447,
+      "step": 5766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.78142523949135e-06,
+      "loss": 2.4863,
+      "step": 5767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.778249151003411e-06,
+      "loss": 2.4902,
+      "step": 5768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7750740862079473e-06,
+      "loss": 2.5599,
+      "step": 5769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7719000456273667e-06,
+      "loss": 2.5201,
+      "step": 5770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7687270297839075e-06,
+      "loss": 2.3115,
+      "step": 5771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.765555039199641e-06,
+      "loss": 2.5017,
+      "step": 5772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.762384074396472e-06,
+      "loss": 2.4697,
+      "step": 5773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.759214135896124e-06,
+      "loss": 2.4611,
+      "step": 5774
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7560452242201695e-06,
+      "loss": 2.5012,
+      "step": 5775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7528773398900065e-06,
+      "loss": 2.5195,
+      "step": 5776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7497104834268515e-06,
+      "loss": 2.4245,
+      "step": 5777
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.746544655351767e-06,
+      "loss": 2.3819,
+      "step": 5778
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7433798561856406e-06,
+      "loss": 2.4622,
+      "step": 5779
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7402160864491887e-06,
+      "loss": 2.5209,
+      "step": 5780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7370533466629665e-06,
+      "loss": 2.4328,
+      "step": 5781
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7338916373473466e-06,
+      "loss": 2.3842,
+      "step": 5782
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.730730959022538e-06,
+      "loss": 2.5227,
+      "step": 5783
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7275713122085923e-06,
+      "loss": 2.5537,
+      "step": 5784
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.724412697425369e-06,
+      "loss": 2.4194,
+      "step": 5785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7212551151925735e-06,
+      "loss": 2.4065,
+      "step": 5786
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7180985660297365e-06,
+      "loss": 2.5087,
+      "step": 5787
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.71494305045622e-06,
+      "loss": 2.4493,
+      "step": 5788
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.711788568991214e-06,
+      "loss": 2.5751,
+      "step": 5789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.708635122153743e-06,
+      "loss": 2.4899,
+      "step": 5790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.705482710462648e-06,
+      "loss": 2.5391,
+      "step": 5791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7023313344366197e-06,
+      "loss": 2.325,
+      "step": 5792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.699180994594167e-06,
+      "loss": 2.524,
+      "step": 5793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6960316914536244e-06,
+      "loss": 2.4399,
+      "step": 5794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6928834255331625e-06,
+      "loss": 2.3528,
+      "step": 5795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6897361973507796e-06,
+      "loss": 2.3459,
+      "step": 5796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6865900074243045e-06,
+      "loss": 2.4572,
+      "step": 5797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.683444856271393e-06,
+      "loss": 2.4166,
+      "step": 5798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680300744409532e-06,
+      "loss": 2.4183,
+      "step": 5799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.677157672356034e-06,
+      "loss": 2.4856,
+      "step": 5800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.674015640628048e-06,
+      "loss": 2.5791,
+      "step": 5801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6708746497425384e-06,
+      "loss": 2.471,
+      "step": 5802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.667734700216312e-06,
+      "loss": 2.4732,
+      "step": 5803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6645957925659967e-06,
+      "loss": 2.4743,
+      "step": 5804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.661457927308052e-06,
+      "loss": 2.4323,
+      "step": 5805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6583211049587643e-06,
+      "loss": 2.546,
+      "step": 5806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.655185326034252e-06,
+      "loss": 2.4695,
+      "step": 5807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.65205059105045e-06,
+      "loss": 2.431,
+      "step": 5808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6489169005231396e-06,
+      "loss": 2.4593,
+      "step": 5809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6457842549679212e-06,
+      "loss": 2.4724,
+      "step": 5810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6426526549002163e-06,
+      "loss": 2.508,
+      "step": 5811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6395221008352844e-06,
+      "loss": 2.4364,
+      "step": 5812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.636392593288208e-06,
+      "loss": 2.5358,
+      "step": 5813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6332641327739017e-06,
+      "loss": 2.4662,
+      "step": 5814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6301367198071035e-06,
+      "loss": 2.4906,
+      "step": 5815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6270103549023804e-06,
+      "loss": 2.4985,
+      "step": 5816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6238850385741277e-06,
+      "loss": 2.4516,
+      "step": 5817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6207607713365713e-06,
+      "loss": 2.4351,
+      "step": 5818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.617637553703753e-06,
+      "loss": 2.444,
+      "step": 5819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6145153861895545e-06,
+      "loss": 2.4057,
+      "step": 5820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6113942693076798e-06,
+      "loss": 2.5469,
+      "step": 5821
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.608274203571659e-06,
+      "loss": 2.4708,
+      "step": 5822
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6051551894948512e-06,
+      "loss": 2.4185,
+      "step": 5823
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.602037227590446e-06,
+      "loss": 2.5232,
+      "step": 5824
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.598920318371444e-06,
+      "loss": 2.4641,
+      "step": 5825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5958044623506948e-06,
+      "loss": 2.4388,
+      "step": 5826
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5926896600408644e-06,
+      "loss": 2.4247,
+      "step": 5827
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5895759119544383e-06,
+      "loss": 2.472,
+      "step": 5828
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5864632186037385e-06,
+      "loss": 2.414,
+      "step": 5829
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.583351580500911e-06,
+      "loss": 2.4922,
+      "step": 5830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.580240998157928e-06,
+      "loss": 2.6433,
+      "step": 5831
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.577131472086585e-06,
+      "loss": 2.4356,
+      "step": 5832
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.574023002798509e-06,
+      "loss": 2.5089,
+      "step": 5833
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5709155908051484e-06,
+      "loss": 2.4795,
+      "step": 5834
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.567809236617784e-06,
+      "loss": 2.448,
+      "step": 5835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5647039407475115e-06,
+      "loss": 2.4452,
+      "step": 5836
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5615997037052576e-06,
+      "loss": 2.4448,
+      "step": 5837
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.558496526001789e-06,
+      "loss": 2.5589,
+      "step": 5838
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5553944081476733e-06,
+      "loss": 2.4981,
+      "step": 5839
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5522933506533196e-06,
+      "loss": 2.401,
+      "step": 5840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5491933540289627e-06,
+      "loss": 2.5748,
+      "step": 5841
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546094418784647e-06,
+      "loss": 2.4912,
+      "step": 5842
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.542996545430267e-06,
+      "loss": 2.5109,
+      "step": 5843
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.539899734475528e-06,
+      "loss": 2.4907,
+      "step": 5844
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5368039864299552e-06,
+      "loss": 2.3975,
+      "step": 5845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.53370930180291e-06,
+      "loss": 2.5514,
+      "step": 5846
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5306156811035753e-06,
+      "loss": 2.4537,
+      "step": 5847
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5275231248409556e-06,
+      "loss": 2.5727,
+      "step": 5848
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.524431633523886e-06,
+      "loss": 2.4397,
+      "step": 5849
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.521341207661022e-06,
+      "loss": 2.4233,
+      "step": 5850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5182518477608463e-06,
+      "loss": 2.3595,
+      "step": 5851
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5151635543316663e-06,
+      "loss": 2.4853,
+      "step": 5852
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.512076327881607e-06,
+      "loss": 2.4212,
+      "step": 5853
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5089901689186247e-06,
+      "loss": 2.4165,
+      "step": 5854
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.505905077950509e-06,
+      "loss": 2.4072,
+      "step": 5855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5028210554848505e-06,
+      "loss": 2.4484,
+      "step": 5856
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4997381020290842e-06,
+      "loss": 2.4741,
+      "step": 5857
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4966562180904607e-06,
+      "loss": 2.354,
+      "step": 5858
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4935754041760563e-06,
+      "loss": 2.4257,
+      "step": 5859
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.490495660792771e-06,
+      "loss": 2.4436,
+      "step": 5860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.487416988447333e-06,
+      "loss": 2.5051,
+      "step": 5861
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4843393876462816e-06,
+      "loss": 2.4742,
+      "step": 5862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4812628588959916e-06,
+      "loss": 2.5091,
+      "step": 5863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.47818740270266e-06,
+      "loss": 2.4377,
+      "step": 5864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4751130195723036e-06,
+      "loss": 2.6393,
+      "step": 5865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472039710010766e-06,
+      "loss": 2.5123,
+      "step": 5866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4689674745237113e-06,
+      "loss": 2.4755,
+      "step": 5867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.465896313616628e-06,
+      "loss": 2.3609,
+      "step": 5868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4628262277948312e-06,
+      "loss": 2.4342,
+      "step": 5869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4597572175634497e-06,
+      "loss": 2.4679,
+      "step": 5870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4566892834274425e-06,
+      "loss": 2.4402,
+      "step": 5871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4536224258915985e-06,
+      "loss": 2.5401,
+      "step": 5872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4505566454605134e-06,
+      "loss": 2.4633,
+      "step": 5873
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4474919426386144e-06,
+      "loss": 2.4868,
+      "step": 5874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.444428317930154e-06,
+      "loss": 2.459,
+      "step": 5875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.441365771839201e-06,
+      "loss": 2.4328,
+      "step": 5876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.438304304869653e-06,
+      "loss": 2.4508,
+      "step": 5877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4352439175252283e-06,
+      "loss": 2.3741,
+      "step": 5878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.432184610309459e-06,
+      "loss": 2.4514,
+      "step": 5879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4291263837257103e-06,
+      "loss": 2.427,
+      "step": 5880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4260692382771667e-06,
+      "loss": 2.4464,
+      "step": 5881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4230131744668327e-06,
+      "loss": 2.3857,
+      "step": 5882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4199581927975377e-06,
+      "loss": 2.5056,
+      "step": 5883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.416904293771931e-06,
+      "loss": 2.494,
+      "step": 5884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4138514778924823e-06,
+      "loss": 2.5771,
+      "step": 5885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4107997456614915e-06,
+      "loss": 2.4953,
+      "step": 5886
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4077490975810656e-06,
+      "loss": 2.517,
+      "step": 5887
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.404699534153141e-06,
+      "loss": 2.45,
+      "step": 5888
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4016510558794857e-06,
+      "loss": 2.406,
+      "step": 5889
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3986036632616703e-06,
+      "loss": 2.4655,
+      "step": 5890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3955573568011e-06,
+      "loss": 2.3745,
+      "step": 5891
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3925121369989954e-06,
+      "loss": 2.4956,
+      "step": 5892
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.389468004356401e-06,
+      "loss": 2.4894,
+      "step": 5893
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3864249593741826e-06,
+      "loss": 2.4997,
+      "step": 5894
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3833830025530266e-06,
+      "loss": 2.4928,
+      "step": 5895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3803421343934326e-06,
+      "loss": 2.5156,
+      "step": 5896
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3773023553957406e-06,
+      "loss": 2.3436,
+      "step": 5897
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3742636660600883e-06,
+      "loss": 2.432,
+      "step": 5898
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3712260668864483e-06,
+      "loss": 2.4221,
+      "step": 5899
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3681895583746116e-06,
+      "loss": 2.441,
+      "step": 5900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3651541410241874e-06,
+      "loss": 2.4676,
+      "step": 5901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3621198153346068e-06,
+      "loss": 2.3363,
+      "step": 5902
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3590865818051244e-06,
+      "loss": 2.489,
+      "step": 5903
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.356054440934804e-06,
+      "loss": 2.4862,
+      "step": 5904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.353023393222539e-06,
+      "loss": 2.4304,
+      "step": 5905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.34999343916705e-06,
+      "loss": 2.5086,
+      "step": 5906
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3469645792668593e-06,
+      "loss": 2.4091,
+      "step": 5907
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3439368140203222e-06,
+      "loss": 2.4079,
+      "step": 5908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3409101439256098e-06,
+      "loss": 2.4737,
+      "step": 5909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.337884569480715e-06,
+      "loss": 2.4569,
+      "step": 5910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3348600911834472e-06,
+      "loss": 2.4569,
+      "step": 5911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.331836709531443e-06,
+      "loss": 2.4234,
+      "step": 5912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.328814425022141e-06,
+      "loss": 2.3167,
+      "step": 5913
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3257932381528245e-06,
+      "loss": 2.4036,
+      "step": 5914
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3227731494205752e-06,
+      "loss": 2.3828,
+      "step": 5915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3197541593223013e-06,
+      "loss": 2.4055,
+      "step": 5916
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3167362683547342e-06,
+      "loss": 2.5286,
+      "step": 5917
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3137194770144186e-06,
+      "loss": 2.5103,
+      "step": 5918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.310703785797723e-06,
+      "loss": 2.4629,
+      "step": 5919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.307689195200834e-06,
+      "loss": 2.462,
+      "step": 5920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.30467570571975e-06,
+      "loss": 2.519,
+      "step": 5921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.301663317850293e-06,
+      "loss": 2.5484,
+      "step": 5922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2986520320881155e-06,
+      "loss": 2.4546,
+      "step": 5923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2956418489286666e-06,
+      "loss": 2.3866,
+      "step": 5924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.29263276886723e-06,
+      "loss": 2.5259,
+      "step": 5925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.289624792398902e-06,
+      "loss": 2.4752,
+      "step": 5926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2866179200186e-06,
+      "loss": 2.5021,
+      "step": 5927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2836121522210564e-06,
+      "loss": 2.4064,
+      "step": 5928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2806074895008276e-06,
+      "loss": 2.3971,
+      "step": 5929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2776039323522734e-06,
+      "loss": 2.4624,
+      "step": 5930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.274601481269598e-06,
+      "loss": 2.4431,
+      "step": 5931
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2716001367467953e-06,
+      "loss": 2.5058,
+      "step": 5932
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2685998992776956e-06,
+      "loss": 2.5122,
+      "step": 5933
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.26560076935594e-06,
+      "loss": 2.4923,
+      "step": 5934
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2626027474749888e-06,
+      "loss": 2.551,
+      "step": 5935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2596058341281212e-06,
+      "loss": 2.4733,
+      "step": 5936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2566100298084334e-06,
+      "loss": 2.4382,
+      "step": 5937
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253615335008835e-06,
+      "loss": 2.5271,
+      "step": 5938
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.250621750222055e-06,
+      "loss": 2.3569,
+      "step": 5939
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2476292759406513e-06,
+      "loss": 2.4522,
+      "step": 5940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244637912656979e-06,
+      "loss": 2.5527,
+      "step": 5941
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241647660863224e-06,
+      "loss": 2.4772,
+      "step": 5942
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2386585210513867e-06,
+      "loss": 2.4652,
+      "step": 5943
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.235670493713281e-06,
+      "loss": 2.4536,
+      "step": 5944
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.232683579340543e-06,
+      "loss": 2.4117,
+      "step": 5945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.229697778424625e-06,
+      "loss": 2.4362,
+      "step": 5946
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2267130914567855e-06,
+      "loss": 2.5413,
+      "step": 5947
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.223729518928117e-06,
+      "loss": 2.4861,
+      "step": 5948
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.220747061329521e-06,
+      "loss": 2.4478,
+      "step": 5949
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2177657191517063e-06,
+      "loss": 2.4706,
+      "step": 5950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214785492885213e-06,
+      "loss": 2.4176,
+      "step": 5951
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.211806383020387e-06,
+      "loss": 2.4113,
+      "step": 5952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.208828390047397e-06,
+      "loss": 2.4162,
+      "step": 5953
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2058515144562295e-06,
+      "loss": 2.4256,
+      "step": 5954
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.202875756736671e-06,
+      "loss": 2.4618,
+      "step": 5955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.199901117378348e-06,
+      "loss": 2.4242,
+      "step": 5956
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1969275968706893e-06,
+      "loss": 2.4766,
+      "step": 5957
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.193955195702937e-06,
+      "loss": 2.4722,
+      "step": 5958
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.190983914364155e-06,
+      "loss": 2.4851,
+      "step": 5959
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.188013753343224e-06,
+      "loss": 2.4837,
+      "step": 5960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1850447131288353e-06,
+      "loss": 2.5111,
+      "step": 5961
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1820767942094987e-06,
+      "loss": 2.4818,
+      "step": 5962
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.179109997073544e-06,
+      "loss": 2.5121,
+      "step": 5963
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1761443222091003e-06,
+      "loss": 2.3313,
+      "step": 5964
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1731797701041343e-06,
+      "loss": 2.4647,
+      "step": 5965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.170216341246416e-06,
+      "loss": 2.5441,
+      "step": 5966
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.167254036123526e-06,
+      "loss": 2.4711,
+      "step": 5967
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.164292855222868e-06,
+      "loss": 2.4954,
+      "step": 5968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1613327990316576e-06,
+      "loss": 2.3962,
+      "step": 5969
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.158373868036928e-06,
+      "loss": 2.3944,
+      "step": 5970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.155416062725528e-06,
+      "loss": 2.5006,
+      "step": 5971
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.152459383584109e-06,
+      "loss": 2.4664,
+      "step": 5972
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.149503831099154e-06,
+      "loss": 2.4859,
+      "step": 5973
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1465494057569567e-06,
+      "loss": 2.3966,
+      "step": 5974
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.143596108043614e-06,
+      "loss": 2.4264,
+      "step": 5975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.140643938445047e-06,
+      "loss": 2.5705,
+      "step": 5976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1376928974469924e-06,
+      "loss": 2.4576,
+      "step": 5977
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1347429855349955e-06,
+      "loss": 2.5013,
+      "step": 5978
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.131794203194419e-06,
+      "loss": 2.4142,
+      "step": 5979
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1288465509104428e-06,
+      "loss": 2.4087,
+      "step": 5980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1259000291680484e-06,
+      "loss": 2.3955,
+      "step": 5981
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1229546384520483e-06,
+      "loss": 2.4848,
+      "step": 5982
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.120010379247063e-06,
+      "loss": 2.4491,
+      "step": 5983
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1170672520375155e-06,
+      "loss": 2.4362,
+      "step": 5984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1141252573076563e-06,
+      "loss": 2.4544,
+      "step": 5985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1111843955415456e-06,
+      "loss": 2.4988,
+      "step": 5986
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1082446672230548e-06,
+      "loss": 2.4593,
+      "step": 5987
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105306072835873e-06,
+      "loss": 2.4202,
+      "step": 5988
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1023686128634978e-06,
+      "loss": 2.4335,
+      "step": 5989
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0994322877892435e-06,
+      "loss": 2.4387,
+      "step": 5990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0964970980962405e-06,
+      "loss": 2.5177,
+      "step": 5991
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0935630442674224e-06,
+      "loss": 2.4441,
+      "step": 5992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.090630126785541e-06,
+      "loss": 2.4391,
+      "step": 5993
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087698346133172e-06,
+      "loss": 2.5023,
+      "step": 5994
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.084767702792687e-06,
+      "loss": 2.4361,
+      "step": 5995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0818381972462785e-06,
+      "loss": 2.4172,
+      "step": 5996
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0789098299759546e-06,
+      "loss": 2.3496,
+      "step": 5997
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0759826014635254e-06,
+      "loss": 2.4752,
+      "step": 5998
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0730565121906274e-06,
+      "loss": 2.4487,
+      "step": 5999
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.070131562638705e-06,
+      "loss": 2.4427,
+      "step": 6000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0672077532890065e-06,
+      "loss": 2.4444,
+      "step": 6001
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.064285084622601e-06,
+      "loss": 2.454,
+      "step": 6002
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0613635571203713e-06,
+      "loss": 2.5372,
+      "step": 6003
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.058443171263008e-06,
+      "loss": 2.4052,
+      "step": 6004
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0555239275310146e-06,
+      "loss": 2.4129,
+      "step": 6005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0526058264047064e-06,
+      "loss": 2.4258,
+      "step": 6006
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0496888683642145e-06,
+      "loss": 2.4448,
+      "step": 6007
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.04677305388948e-06,
+      "loss": 2.5249,
+      "step": 6008
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0438583834602486e-06,
+      "loss": 2.4578,
+      "step": 6009
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.040944857556085e-06,
+      "loss": 2.431,
+      "step": 6010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.038032476656374e-06,
+      "loss": 2.5401,
+      "step": 6011
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.035121241240293e-06,
+      "loss": 2.4993,
+      "step": 6012
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0322111517868435e-06,
+      "loss": 2.4343,
+      "step": 6013
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.029302208774836e-06,
+      "loss": 2.5024,
+      "step": 6014
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0263944126828902e-06,
+      "loss": 2.497,
+      "step": 6015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.023487763989442e-06,
+      "loss": 2.4171,
+      "step": 6016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0205822631727355e-06,
+      "loss": 2.4537,
+      "step": 6017
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017677910710821e-06,
+      "loss": 2.3224,
+      "step": 6018
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.014774707081567e-06,
+      "loss": 2.4131,
+      "step": 6019
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.011872652762651e-06,
+      "loss": 2.438,
+      "step": 6020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0089717482315605e-06,
+      "loss": 2.4052,
+      "step": 6021
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0060719939655945e-06,
+      "loss": 2.4152,
+      "step": 6022
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0031733904418625e-06,
+      "loss": 2.4889,
+      "step": 6023
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.000275938137286e-06,
+      "loss": 2.4134,
+      "step": 6024
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997379637528597e-06,
+      "loss": 2.5188,
+      "step": 6025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9944844890923307e-06,
+      "loss": 2.4076,
+      "step": 6026
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.99159049330484e-06,
+      "loss": 2.4229,
+      "step": 6027
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.988697650642296e-06,
+      "loss": 2.5305,
+      "step": 6028
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.98580596158066e-06,
+      "loss": 2.4991,
+      "step": 6029
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9829154265957205e-06,
+      "loss": 2.4832,
+      "step": 6030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.980026046163068e-06,
+      "loss": 2.5123,
+      "step": 6031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9771378207581057e-06,
+      "loss": 2.4263,
+      "step": 6032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.974250750856047e-06,
+      "loss": 2.5053,
+      "step": 6033
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.971364836931917e-06,
+      "loss": 2.3064,
+      "step": 6034
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9684800794605406e-06,
+      "loss": 2.5041,
+      "step": 6035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9655964789165648e-06,
+      "loss": 2.3542,
+      "step": 6036
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.96271403577444e-06,
+      "loss": 2.4237,
+      "step": 6037
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.959832750508429e-06,
+      "loss": 2.4071,
+      "step": 6038
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9569526235926003e-06,
+      "loss": 2.4716,
+      "step": 6039
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9540736555008367e-06,
+      "loss": 2.5306,
+      "step": 6040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9511958467068246e-06,
+      "loss": 2.5573,
+      "step": 6041
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9483191976840688e-06,
+      "loss": 2.4534,
+      "step": 6042
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9454437089058695e-06,
+      "loss": 2.4745,
+      "step": 6043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.942569380845344e-06,
+      "loss": 2.5063,
+      "step": 6044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.939696213975427e-06,
+      "loss": 2.5147,
+      "step": 6045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9368242087688447e-06,
+      "loss": 2.4699,
+      "step": 6046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9339533656981444e-06,
+      "loss": 2.3694,
+      "step": 6047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9310836852356782e-06,
+      "loss": 2.4033,
+      "step": 6048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9282151678536087e-06,
+      "loss": 2.5019,
+      "step": 6049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9253478140239043e-06,
+      "loss": 2.4194,
+      "step": 6050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9224816242183475e-06,
+      "loss": 2.4517,
+      "step": 6051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9196165989085155e-06,
+      "loss": 2.5662,
+      "step": 6052
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9167527385658155e-06,
+      "loss": 2.3473,
+      "step": 6053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.913890043661445e-06,
+      "loss": 2.4597,
+      "step": 6054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.911028514666415e-06,
+      "loss": 2.4476,
+      "step": 6055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.908168152051548e-06,
+      "loss": 2.5257,
+      "step": 6056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9053089562874726e-06,
+      "loss": 2.4742,
+      "step": 6057
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9024509278446233e-06,
+      "loss": 2.5248,
+      "step": 6058
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.899594067193249e-06,
+      "loss": 2.4798,
+      "step": 6059
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.896738374803395e-06,
+      "loss": 2.4028,
+      "step": 6060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.893883851144921e-06,
+      "loss": 2.4093,
+      "step": 6061
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8910304966875036e-06,
+      "loss": 2.4122,
+      "step": 6062
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.888178311900609e-06,
+      "loss": 2.4225,
+      "step": 6063
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.885327297253523e-06,
+      "loss": 2.4817,
+      "step": 6064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8824774532153355e-06,
+      "loss": 2.4627,
+      "step": 6065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8796287802549448e-06,
+      "loss": 2.4263,
+      "step": 6066
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8767812788410544e-06,
+      "loss": 2.436,
+      "step": 6067
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.873934949442182e-06,
+      "loss": 2.4839,
+      "step": 6068
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8710897925266344e-06,
+      "loss": 2.5012,
+      "step": 6069
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.868245808562553e-06,
+      "loss": 2.4618,
+      "step": 6070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.865402998017861e-06,
+      "loss": 2.5412,
+      "step": 6071
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.862561361360301e-06,
+      "loss": 2.4292,
+      "step": 6072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.859720899057422e-06,
+      "loss": 2.4123,
+      "step": 6073
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.856881611576576e-06,
+      "loss": 2.4271,
+      "step": 6074
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8540434993849253e-06,
+      "loss": 2.4355,
+      "step": 6075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.851206562949439e-06,
+      "loss": 2.4816,
+      "step": 6076
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.848370802736885e-06,
+      "loss": 2.4437,
+      "step": 6077
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.845536219213845e-06,
+      "loss": 2.481,
+      "step": 6078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8427028128467127e-06,
+      "loss": 2.4445,
+      "step": 6079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.839870584101673e-06,
+      "loss": 2.4586,
+      "step": 6080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8370395334447286e-06,
+      "loss": 2.3885,
+      "step": 6081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8342096613416847e-06,
+      "loss": 2.508,
+      "step": 6082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.831380968258153e-06,
+      "loss": 2.4338,
+      "step": 6083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8285534546595497e-06,
+      "loss": 2.3912,
+      "step": 6084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8257271210111024e-06,
+      "loss": 2.3947,
+      "step": 6085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8229019677778314e-06,
+      "loss": 2.4499,
+      "step": 6086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.820077995424584e-06,
+      "loss": 2.4422,
+      "step": 6087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8172552044159917e-06,
+      "loss": 2.4605,
+      "step": 6088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8144335952165034e-06,
+      "loss": 2.3959,
+      "step": 6089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.811613168290371e-06,
+      "loss": 2.5166,
+      "step": 6090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8087939241016538e-06,
+      "loss": 2.4558,
+      "step": 6091
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.805975863114212e-06,
+      "loss": 2.4123,
+      "step": 6092
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8031589857917186e-06,
+      "loss": 2.5113,
+      "step": 6093
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8003432925976402e-06,
+      "loss": 2.431,
+      "step": 6094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.797528783995257e-06,
+      "loss": 2.3706,
+      "step": 6095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.794715460447659e-06,
+      "loss": 2.4795,
+      "step": 6096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7919033224177273e-06,
+      "loss": 2.4756,
+      "step": 6097
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7890923703681592e-06,
+      "loss": 2.5413,
+      "step": 6098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786282604761452e-06,
+      "loss": 2.5101,
+      "step": 6099
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.783474026059909e-06,
+      "loss": 2.4981,
+      "step": 6100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7806666347256394e-06,
+      "loss": 2.4688,
+      "step": 6101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7778604312205583e-06,
+      "loss": 2.488,
+      "step": 6102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7750554160063726e-06,
+      "loss": 2.4581,
+      "step": 6103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7722515895446178e-06,
+      "loss": 2.4985,
+      "step": 6104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7694489522966104e-06,
+      "loss": 2.5327,
+      "step": 6105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.766647504723484e-06,
+      "loss": 2.4099,
+      "step": 6106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.763847247286173e-06,
+      "loss": 2.4986,
+      "step": 6107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.761048180445416e-06,
+      "loss": 2.4119,
+      "step": 6108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.758250304661757e-06,
+      "loss": 2.5432,
+      "step": 6109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.755453620395544e-06,
+      "loss": 2.4715,
+      "step": 6110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.752658128106921e-06,
+      "loss": 2.4066,
+      "step": 6111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.749863828255851e-06,
+      "loss": 2.5696,
+      "step": 6112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.747070721302093e-06,
+      "loss": 2.5175,
+      "step": 6113
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.744278807705203e-06,
+      "loss": 2.4507,
+      "step": 6114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.741488087924551e-06,
+      "loss": 2.4029,
+      "step": 6115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.738698562419304e-06,
+      "loss": 2.5176,
+      "step": 6116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7359102316484375e-06,
+      "loss": 2.3876,
+      "step": 6117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7331230960707277e-06,
+      "loss": 2.3676,
+      "step": 6118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.730337156144758e-06,
+      "loss": 2.497,
+      "step": 6119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7275524123289e-06,
+      "loss": 2.4386,
+      "step": 6120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.724768865081355e-06,
+      "loss": 2.4599,
+      "step": 6121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.721986514860101e-06,
+      "loss": 2.509,
+      "step": 6122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7192053621229366e-06,
+      "loss": 2.3609,
+      "step": 6123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.716425407327453e-06,
+      "loss": 2.4018,
+      "step": 6124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7136466509310533e-06,
+      "loss": 2.5513,
+      "step": 6125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7108690933909355e-06,
+      "loss": 2.461,
+      "step": 6126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7080927351641062e-06,
+      "loss": 2.4886,
+      "step": 6127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7053175767073657e-06,
+      "loss": 2.4946,
+      "step": 6128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.70254361847733e-06,
+      "loss": 2.455,
+      "step": 6129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.699770860930413e-06,
+      "loss": 2.4253,
+      "step": 6130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6969993045228203e-06,
+      "loss": 2.4782,
+      "step": 6131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6942289497105732e-06,
+      "loss": 2.4993,
+      "step": 6132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6914597969494903e-06,
+      "loss": 2.465,
+      "step": 6133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6886918466951926e-06,
+      "loss": 2.4025,
+      "step": 6134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6859250994031026e-06,
+      "loss": 2.5042,
+      "step": 6135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6831595555284494e-06,
+      "loss": 2.4011,
+      "step": 6136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6803952155262513e-06,
+      "loss": 2.4698,
+      "step": 6137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6776320798513456e-06,
+      "loss": 2.3936,
+      "step": 6138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674870148958365e-06,
+      "loss": 2.4307,
+      "step": 6139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.672109423301734e-06,
+      "loss": 2.4498,
+      "step": 6140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.669349903335693e-06,
+      "loss": 2.5084,
+      "step": 6141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.666591589514276e-06,
+      "loss": 2.5003,
+      "step": 6142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.663834482291322e-06,
+      "loss": 2.4817,
+      "step": 6143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.661078582120472e-06,
+      "loss": 2.4612,
+      "step": 6144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6583238894551576e-06,
+      "loss": 2.4602,
+      "step": 6145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.655570404748631e-06,
+      "loss": 2.385,
+      "step": 6146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6528181284539346e-06,
+      "loss": 2.4111,
+      "step": 6147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6500670610239063e-06,
+      "loss": 2.4991,
+      "step": 6148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6473172029111914e-06,
+      "loss": 2.4292,
+      "step": 6149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6445685545682466e-06,
+      "loss": 2.534,
+      "step": 6150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6418211164473094e-06,
+      "loss": 2.3877,
+      "step": 6151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6390748890004313e-06,
+      "loss": 2.4139,
+      "step": 6152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.636329872679465e-06,
+      "loss": 2.4133,
+      "step": 6153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6335860679360494e-06,
+      "loss": 2.4647,
+      "step": 6154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.630843475221646e-06,
+      "loss": 2.5559,
+      "step": 6155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6281020949875045e-06,
+      "loss": 2.4473,
+      "step": 6156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.625361927684671e-06,
+      "loss": 2.411,
+      "step": 6157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6226229737640007e-06,
+      "loss": 2.4242,
+      "step": 6158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6198852336761447e-06,
+      "loss": 2.4118,
+      "step": 6159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6171487078715576e-06,
+      "loss": 2.3517,
+      "step": 6160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6144133968004938e-06,
+      "loss": 2.4313,
+      "step": 6161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.611679300912998e-06,
+      "loss": 2.4969,
+      "step": 6162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6089464206589323e-06,
+      "loss": 2.4243,
+      "step": 6163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.606214756487948e-06,
+      "loss": 2.5716,
+      "step": 6164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.603484308849493e-06,
+      "loss": 2.5085,
+      "step": 6165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6007550781928216e-06,
+      "loss": 2.4557,
+      "step": 6166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598027064966994e-06,
+      "loss": 2.5567,
+      "step": 6167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.595300269620853e-06,
+      "loss": 2.4384,
+      "step": 6168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5925746926030536e-06,
+      "loss": 2.4225,
+      "step": 6169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5898503343620473e-06,
+      "loss": 2.4563,
+      "step": 6170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5871271953460865e-06,
+      "loss": 2.4672,
+      "step": 6171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5844052760032177e-06,
+      "loss": 2.4839,
+      "step": 6172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5816845767812983e-06,
+      "loss": 2.4822,
+      "step": 6173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.578965098127967e-06,
+      "loss": 2.4788,
+      "step": 6174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.576246840490677e-06,
+      "loss": 2.4575,
+      "step": 6175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5735298043166744e-06,
+      "loss": 2.4649,
+      "step": 6176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.570813990053005e-06,
+      "loss": 2.4293,
+      "step": 6177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5680993981465187e-06,
+      "loss": 2.5125,
+      "step": 6178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.565386029043848e-06,
+      "loss": 2.4588,
+      "step": 6179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5626738831914467e-06,
+      "loss": 2.4612,
+      "step": 6180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5599629610355557e-06,
+      "loss": 2.4611,
+      "step": 6181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5572532630222092e-06,
+      "loss": 2.4873,
+      "step": 6182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5545447895972442e-06,
+      "loss": 2.4212,
+      "step": 6183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.551837541206309e-06,
+      "loss": 2.5857,
+      "step": 6184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5491315182948306e-06,
+      "loss": 2.4886,
+      "step": 6185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.546426721308044e-06,
+      "loss": 2.4075,
+      "step": 6186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.543723150690983e-06,
+      "loss": 2.4316,
+      "step": 6187
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.541020806888478e-06,
+      "loss": 2.5015,
+      "step": 6188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.538319690345157e-06,
+      "loss": 2.4784,
+      "step": 6189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.53561980150545e-06,
+      "loss": 2.4487,
+      "step": 6190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532921140813577e-06,
+      "loss": 2.4411,
+      "step": 6191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5302237087135617e-06,
+      "loss": 2.5109,
+      "step": 6192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5275275056492254e-06,
+      "loss": 2.4733,
+      "step": 6193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5248325320641875e-06,
+      "loss": 2.4023,
+      "step": 6194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5221387884018634e-06,
+      "loss": 2.4616,
+      "step": 6195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5194462751054673e-06,
+      "loss": 2.5048,
+      "step": 6196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.516754992618009e-06,
+      "loss": 2.4943,
+      "step": 6197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5140649413823015e-06,
+      "loss": 2.4892,
+      "step": 6198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.511376121840944e-06,
+      "loss": 2.4441,
+      "step": 6199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.508688534436341e-06,
+      "loss": 2.4391,
+      "step": 6200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5060021796107004e-06,
+      "loss": 2.4419,
+      "step": 6201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503317057806014e-06,
+      "loss": 2.3638,
+      "step": 6202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5006331694640785e-06,
+      "loss": 2.4719,
+      "step": 6203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.497950515026485e-06,
+      "loss": 2.4562,
+      "step": 6204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4952690949346246e-06,
+      "loss": 2.3639,
+      "step": 6205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4925889096296817e-06,
+      "loss": 2.3972,
+      "step": 6206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.489909959552643e-06,
+      "loss": 2.4211,
+      "step": 6207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4872322451442786e-06,
+      "loss": 2.3543,
+      "step": 6208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.484555766845177e-06,
+      "loss": 2.3747,
+      "step": 6209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4818805250957036e-06,
+      "loss": 2.4422,
+      "step": 6210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4792065203360284e-06,
+      "loss": 2.3715,
+      "step": 6211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4765337530061184e-06,
+      "loss": 2.3991,
+      "step": 6212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4738622235457366e-06,
+      "loss": 2.3673,
+      "step": 6213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.471191932394441e-06,
+      "loss": 2.4649,
+      "step": 6214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4685228799915904e-06,
+      "loss": 2.468,
+      "step": 6215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.465855066776328e-06,
+      "loss": 2.5966,
+      "step": 6216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.463188493187604e-06,
+      "loss": 2.3922,
+      "step": 6217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4605231596641664e-06,
+      "loss": 2.573,
+      "step": 6218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.45785906664455e-06,
+      "loss": 2.5154,
+      "step": 6219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4551962145670905e-06,
+      "loss": 2.4223,
+      "step": 6220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.452534603869918e-06,
+      "loss": 2.4933,
+      "step": 6221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.449874234990962e-06,
+      "loss": 2.5103,
+      "step": 6222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.447215108367943e-06,
+      "loss": 2.3542,
+      "step": 6223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.444557224438382e-06,
+      "loss": 2.5223,
+      "step": 6224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4419005836395835e-06,
+      "loss": 2.5003,
+      "step": 6225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4392451864086698e-06,
+      "loss": 2.3875,
+      "step": 6226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.436591033182535e-06,
+      "loss": 2.4881,
+      "step": 6227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.433938124397882e-06,
+      "loss": 2.4095,
+      "step": 6228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4312864604912058e-06,
+      "loss": 2.4315,
+      "step": 6229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4286360418987965e-06,
+      "loss": 2.4864,
+      "step": 6230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4259868690567388e-06,
+      "loss": 2.4191,
+      "step": 6231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.423338942400917e-06,
+      "loss": 2.493,
+      "step": 6232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4206922623670004e-06,
+      "loss": 2.5116,
+      "step": 6233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.418046829390458e-06,
+      "loss": 2.4472,
+      "step": 6234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.415402643906564e-06,
+      "loss": 2.5055,
+      "step": 6235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4127597063503694e-06,
+      "loss": 2.4274,
+      "step": 6236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4101180171567316e-06,
+      "loss": 2.402,
+      "step": 6237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4074775767602987e-06,
+      "loss": 2.5043,
+      "step": 6238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.404838385595516e-06,
+      "loss": 2.4134,
+      "step": 6239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4022004440966198e-06,
+      "loss": 2.4045,
+      "step": 6240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3995637526976444e-06,
+      "loss": 2.5174,
+      "step": 6241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3969283118324104e-06,
+      "loss": 2.4816,
+      "step": 6242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3942941219345495e-06,
+      "loss": 2.4999,
+      "step": 6243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.391661183437466e-06,
+      "loss": 2.4565,
+      "step": 6244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3890294967743742e-06,
+      "loss": 2.4495,
+      "step": 6245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3863990623782764e-06,
+      "loss": 2.3404,
+      "step": 6246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.38376988068197e-06,
+      "loss": 2.5272,
+      "step": 6247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3811419521180447e-06,
+      "loss": 2.3864,
+      "step": 6248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.37851527711889e-06,
+      "loss": 2.4985,
+      "step": 6249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.375889856116679e-06,
+      "loss": 2.4974,
+      "step": 6250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3732656895433815e-06,
+      "loss": 2.5251,
+      "step": 6251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.370642777830774e-06,
+      "loss": 2.4607,
+      "step": 6252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3680211214104064e-06,
+      "loss": 2.4698,
+      "step": 6253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3654007207136353e-06,
+      "loss": 2.445,
+      "step": 6254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3627815761716065e-06,
+      "loss": 2.4987,
+      "step": 6255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360163688215259e-06,
+      "loss": 2.4434,
+      "step": 6256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3575470572753266e-06,
+      "loss": 2.4936,
+      "step": 6257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3549316837823378e-06,
+      "loss": 2.3683,
+      "step": 6258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352317568166602e-06,
+      "loss": 2.4084,
+      "step": 6259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.349704710858244e-06,
+      "loss": 2.5216,
+      "step": 6260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3470931122871598e-06,
+      "loss": 2.3705,
+      "step": 6261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.344482772883051e-06,
+      "loss": 2.4358,
+      "step": 6262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3418736930754073e-06,
+      "loss": 2.4961,
+      "step": 6263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.339265873293514e-06,
+      "loss": 2.4773,
+      "step": 6264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.336659313966444e-06,
+      "loss": 2.4912,
+      "step": 6265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.334054015523073e-06,
+      "loss": 2.4045,
+      "step": 6266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.331449978392051e-06,
+      "loss": 2.4994,
+      "step": 6267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3288472030018417e-06,
+      "loss": 2.4014,
+      "step": 6268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3262456897806907e-06,
+      "loss": 2.3935,
+      "step": 6269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323645439156632e-06,
+      "loss": 2.4321,
+      "step": 6270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3210464515574984e-06,
+      "loss": 2.4882,
+      "step": 6271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3184487274109126e-06,
+      "loss": 2.3912,
+      "step": 6272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3158522671442907e-06,
+      "loss": 2.4095,
+      "step": 6273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.31325707118484e-06,
+      "loss": 2.4721,
+      "step": 6274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3106631399595612e-06,
+      "loss": 2.4305,
+      "step": 6275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3080704738952388e-06,
+      "loss": 2.5328,
+      "step": 6276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.305479073418465e-06,
+      "loss": 2.478,
+      "step": 6277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3028889389556077e-06,
+      "loss": 2.4157,
+      "step": 6278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.300300070932835e-06,
+      "loss": 2.3902,
+      "step": 6279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2977124697761056e-06,
+      "loss": 2.4234,
+      "step": 6280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2951261359111676e-06,
+      "loss": 2.4023,
+      "step": 6281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2925410697635653e-06,
+      "loss": 2.5486,
+      "step": 6282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2899572717586304e-06,
+      "loss": 2.4855,
+      "step": 6283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2873747423214807e-06,
+      "loss": 2.5136,
+      "step": 6284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.284793481877039e-06,
+      "loss": 2.445,
+      "step": 6285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2822134908500115e-06,
+      "loss": 2.495,
+      "step": 6286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2796347696648903e-06,
+      "loss": 2.5218,
+      "step": 6287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2770573187459664e-06,
+      "loss": 2.4223,
+      "step": 6288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2744811385173194e-06,
+      "loss": 2.4508,
+      "step": 6289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2719062294028203e-06,
+      "loss": 2.474,
+      "step": 6290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.269332591826129e-06,
+      "loss": 2.4875,
+      "step": 6291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2667602262107015e-06,
+      "loss": 2.5204,
+      "step": 6292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2641891329797726e-06,
+      "loss": 2.4598,
+      "step": 6293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2616193125563857e-06,
+      "loss": 2.5249,
+      "step": 6294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2590507653633574e-06,
+      "loss": 2.3951,
+      "step": 6295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2564834918233057e-06,
+      "loss": 2.4539,
+      "step": 6296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.253917492358634e-06,
+      "loss": 2.4897,
+      "step": 6297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.251352767391538e-06,
+      "loss": 2.5054,
+      "step": 6298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2487893173440034e-06,
+      "loss": 2.4438,
+      "step": 6299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.24622714263781e-06,
+      "loss": 2.4651,
+      "step": 6300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2436662436945143e-06,
+      "loss": 2.4968,
+      "step": 6301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2411066209354813e-06,
+      "loss": 2.4842,
+      "step": 6302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2385482747818565e-06,
+      "loss": 2.4097,
+      "step": 6303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.235991205654572e-06,
+      "loss": 2.4167,
+      "step": 6304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.233435413974352e-06,
+      "loss": 2.4998,
+      "step": 6305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2308809001617217e-06,
+      "loss": 2.4016,
+      "step": 6306
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.228327664636978e-06,
+      "loss": 2.5084,
+      "step": 6307
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2257757078202193e-06,
+      "loss": 2.505,
+      "step": 6308
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.223225030131333e-06,
+      "loss": 2.4343,
+      "step": 6309
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2206756319899846e-06,
+      "loss": 2.416,
+      "step": 6310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.218127513815651e-06,
+      "loss": 2.4184,
+      "step": 6311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2155806760275745e-06,
+      "loss": 2.4778,
+      "step": 6312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.213035119044802e-06,
+      "loss": 2.4163,
+      "step": 6313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2104908432861648e-06,
+      "loss": 2.4096,
+      "step": 6314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2079478491702853e-06,
+      "loss": 2.4628,
+      "step": 6315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2054061371155724e-06,
+      "loss": 2.4648,
+      "step": 6316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2028657075402282e-06,
+      "loss": 2.3844,
+      "step": 6317
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2003265608622327e-06,
+      "loss": 2.3887,
+      "step": 6318
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.197788697499371e-06,
+      "loss": 2.3853,
+      "step": 6319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.195252117869211e-06,
+      "loss": 2.4654,
+      "step": 6320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1927168223890994e-06,
+      "loss": 2.4525,
+      "step": 6321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1901828114761814e-06,
+      "loss": 2.4396,
+      "step": 6322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.187650085547396e-06,
+      "loss": 2.4341,
+      "step": 6323
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1851186450194574e-06,
+      "loss": 2.4828,
+      "step": 6324
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1825884903088757e-06,
+      "loss": 2.4603,
+      "step": 6325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1800596218319503e-06,
+      "loss": 2.4262,
+      "step": 6326
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1775320400047663e-06,
+      "loss": 2.5128,
+      "step": 6327
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1750057452432017e-06,
+      "loss": 2.4111,
+      "step": 6328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.172480737962912e-06,
+      "loss": 2.4551,
+      "step": 6329
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1699570185793516e-06,
+      "loss": 2.4578,
+      "step": 6330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1674345875077585e-06,
+      "loss": 2.375,
+      "step": 6331
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1649134451631616e-06,
+      "loss": 2.4249,
+      "step": 6332
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1623935919603736e-06,
+      "loss": 2.4378,
+      "step": 6333
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1598750283140014e-06,
+      "loss": 2.5442,
+      "step": 6334
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1573577546384263e-06,
+      "loss": 2.351,
+      "step": 6335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1548417713478355e-06,
+      "loss": 2.4953,
+      "step": 6336
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1523270788561935e-06,
+      "loss": 2.5194,
+      "step": 6337
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1498136775772505e-06,
+      "loss": 2.5016,
+      "step": 6338
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1473015679245467e-06,
+      "loss": 2.441,
+      "step": 6339
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1447907503114196e-06,
+      "loss": 2.4578,
+      "step": 6340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.142281225150975e-06,
+      "loss": 2.4188,
+      "step": 6341
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1397729928561217e-06,
+      "loss": 2.4878,
+      "step": 6342
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1372660538395484e-06,
+      "loss": 2.3895,
+      "step": 6343
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.134760408513734e-06,
+      "loss": 2.527,
+      "step": 6344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1322560572909423e-06,
+      "loss": 2.4102,
+      "step": 6345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.12975300058323e-06,
+      "loss": 2.4033,
+      "step": 6346
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.127251238802429e-06,
+      "loss": 2.426,
+      "step": 6347
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.12475077236017e-06,
+      "loss": 2.4343,
+      "step": 6348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122251601667864e-06,
+      "loss": 2.5008,
+      "step": 6349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.119753727136712e-06,
+      "loss": 2.5037,
+      "step": 6350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1172571491777028e-06,
+      "loss": 2.4158,
+      "step": 6351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1147618682016004e-06,
+      "loss": 2.5288,
+      "step": 6352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1122678846189736e-06,
+      "loss": 2.4798,
+      "step": 6353
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.109775198840169e-06,
+      "loss": 2.4713,
+      "step": 6354
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1072838112753136e-06,
+      "loss": 2.437,
+      "step": 6355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.104793722334326e-06,
+      "loss": 2.5015,
+      "step": 6356
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1023049324269206e-06,
+      "loss": 2.5812,
+      "step": 6357
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0998174419625804e-06,
+      "loss": 2.4902,
+      "step": 6358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.097331251350587e-06,
+      "loss": 2.4659,
+      "step": 6359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0948463610000036e-06,
+      "loss": 2.3892,
+      "step": 6360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0923627713196794e-06,
+      "loss": 2.3243,
+      "step": 6361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.089880482718253e-06,
+      "loss": 2.5004,
+      "step": 6362
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0873994956041468e-06,
+      "loss": 2.4578,
+      "step": 6363
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0849198103855607e-06,
+      "loss": 2.3626,
+      "step": 6364
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0824414274705007e-06,
+      "loss": 2.3949,
+      "step": 6365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0799643472667374e-06,
+      "loss": 2.4762,
+      "step": 6366
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.077488570181839e-06,
+      "loss": 2.4573,
+      "step": 6367
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.075014096623158e-06,
+      "loss": 2.4174,
+      "step": 6368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0725409269978235e-06,
+      "loss": 2.5019,
+      "step": 6369
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.070069061712764e-06,
+      "loss": 2.4226,
+      "step": 6370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.067598501174687e-06,
+      "loss": 2.4322,
+      "step": 6371
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0651292457900806e-06,
+      "loss": 2.4342,
+      "step": 6372
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0626612959652216e-06,
+      "loss": 2.3901,
+      "step": 6373
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.06019465210618e-06,
+      "loss": 2.4357,
+      "step": 6374
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.057729314618796e-06,
+      "loss": 2.5355,
+      "step": 6375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.055265283908707e-06,
+      "loss": 2.4813,
+      "step": 6376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0528025603813296e-06,
+      "loss": 2.5642,
+      "step": 6377
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0503411444418664e-06,
+      "loss": 2.4941,
+      "step": 6378
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047881036495306e-06,
+      "loss": 2.4767,
+      "step": 6379
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0454222369464226e-06,
+      "loss": 2.4576,
+      "step": 6380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.042964746199767e-06,
+      "loss": 2.4136,
+      "step": 6381
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.04050856465969e-06,
+      "loss": 2.4301,
+      "step": 6382
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.038053692730312e-06,
+      "loss": 2.4492,
+      "step": 6383
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035600130815545e-06,
+      "loss": 2.4398,
+      "step": 6384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033147879319086e-06,
+      "loss": 2.3686,
+      "step": 6385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.030696938644414e-06,
+      "loss": 2.4392,
+      "step": 6386
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.028247309194793e-06,
+      "loss": 2.5021,
+      "step": 6387
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0257989913732747e-06,
+      "loss": 2.4106,
+      "step": 6388
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0233519855826864e-06,
+      "loss": 2.4678,
+      "step": 6389
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.020906292225645e-06,
+      "loss": 2.4481,
+      "step": 6390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.018461911704559e-06,
+      "loss": 2.4886,
+      "step": 6391
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0160188444216056e-06,
+      "loss": 2.4387,
+      "step": 6392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0135770907787555e-06,
+      "loss": 2.3525,
+      "step": 6393
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.011136651177763e-06,
+      "loss": 2.563,
+      "step": 6394
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0086975260201624e-06,
+      "loss": 2.4804,
+      "step": 6395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.006259715707275e-06,
+      "loss": 2.5683,
+      "step": 6396
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0038232206402076e-06,
+      "loss": 2.4397,
+      "step": 6397
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0013880412198382e-06,
+      "loss": 2.5209,
+      "step": 6398
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.998954177846851e-06,
+      "loss": 2.4348,
+      "step": 6399
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9965216309216893e-06,
+      "loss": 2.3865,
+      "step": 6400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.994090400844596e-06,
+      "loss": 2.5418,
+      "step": 6401
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.991660488015592e-06,
+      "loss": 2.357,
+      "step": 6402
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98923189283448e-06,
+      "loss": 2.3549,
+      "step": 6403
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.986804615700849e-06,
+      "loss": 2.4563,
+      "step": 6404
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.984378657014072e-06,
+      "loss": 2.424,
+      "step": 6405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.981954017173299e-06,
+      "loss": 2.46,
+      "step": 6406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.979530696577464e-06,
+      "loss": 2.417,
+      "step": 6407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9771086956252962e-06,
+      "loss": 2.4226,
+      "step": 6408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9746880147152912e-06,
+      "loss": 2.4627,
+      "step": 6409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.972268654245736e-06,
+      "loss": 2.4682,
+      "step": 6410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.969850614614698e-06,
+      "loss": 2.443,
+      "step": 6411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.96743389622003e-06,
+      "loss": 2.4965,
+      "step": 6412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9650184994593647e-06,
+      "loss": 2.514,
+      "step": 6413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.96260442473012e-06,
+      "loss": 2.5308,
+      "step": 6414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9601916724294865e-06,
+      "loss": 2.4095,
+      "step": 6415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9577802429544577e-06,
+      "loss": 2.5322,
+      "step": 6416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9553701367017863e-06,
+      "loss": 2.472,
+      "step": 6417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9529613540680226e-06,
+      "loss": 2.462,
+      "step": 6418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.950553895449493e-06,
+      "loss": 2.5038,
+      "step": 6419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.948147761242308e-06,
+      "loss": 2.3748,
+      "step": 6420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9457429518423598e-06,
+      "loss": 2.4587,
+      "step": 6421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.943339467645324e-06,
+      "loss": 2.5636,
+      "step": 6422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.940937309046651e-06,
+      "loss": 2.405,
+      "step": 6423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9385364764415836e-06,
+      "loss": 2.4536,
+      "step": 6424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936136970225143e-06,
+      "loss": 2.5086,
+      "step": 6425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9337387907921267e-06,
+      "loss": 2.4405,
+      "step": 6426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.931341938537118e-06,
+      "loss": 2.4699,
+      "step": 6427
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9289464138544834e-06,
+      "loss": 2.4281,
+      "step": 6428
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.926552217138369e-06,
+      "loss": 2.3938,
+      "step": 6429
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9241593487827026e-06,
+      "loss": 2.5448,
+      "step": 6430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9217678091811964e-06,
+      "loss": 2.5227,
+      "step": 6431
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.919377598727332e-06,
+      "loss": 2.4485,
+      "step": 6432
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916988717814393e-06,
+      "loss": 2.5032,
+      "step": 6433
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914601166835425e-06,
+      "loss": 2.415,
+      "step": 6434
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.912214946183264e-06,
+      "loss": 2.4843,
+      "step": 6435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 2.518,
+      "step": 6436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9074464974296093e-06,
+      "loss": 2.4421,
+      "step": 6437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9050642701126886e-06,
+      "loss": 2.3822,
+      "step": 6438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9026833746917261e-06,
+      "loss": 2.45,
+      "step": 6439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.900303811558455e-06,
+      "loss": 2.4168,
+      "step": 6440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8979255811044006e-06,
+      "loss": 2.4444,
+      "step": 6441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.895548683720866e-06,
+      "loss": 2.4599,
+      "step": 6442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8931731197989267e-06,
+      "loss": 2.4463,
+      "step": 6443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.890798889729447e-06,
+      "loss": 2.5473,
+      "step": 6444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8884259939030703e-06,
+      "loss": 2.5615,
+      "step": 6445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8860544327102182e-06,
+      "loss": 2.4639,
+      "step": 6446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8836842065410965e-06,
+      "loss": 2.4582,
+      "step": 6447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.881315315785691e-06,
+      "loss": 2.4423,
+      "step": 6448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.878947760833757e-06,
+      "loss": 2.4838,
+      "step": 6449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8765815420748512e-06,
+      "loss": 2.5265,
+      "step": 6450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8742166598982881e-06,
+      "loss": 2.5065,
+      "step": 6451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8718531146931762e-06,
+      "loss": 2.4412,
+      "step": 6452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8694909068484002e-06,
+      "loss": 2.4823,
+      "step": 6453
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8671300367526245e-06,
+      "loss": 2.4063,
+      "step": 6454
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8647705047942944e-06,
+      "loss": 2.4579,
+      "step": 6455
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8624123113616355e-06,
+      "loss": 2.4453,
+      "step": 6456
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8600554568426442e-06,
+      "loss": 2.4096,
+      "step": 6457
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.857699941625113e-06,
+      "loss": 2.5068,
+      "step": 6458
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8553457660966046e-06,
+      "loss": 2.4808,
+      "step": 6459
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8529929306444583e-06,
+      "loss": 2.447,
+      "step": 6460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8506414356557966e-06,
+      "loss": 2.5048,
+      "step": 6461
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8482912815175236e-06,
+      "loss": 2.351,
+      "step": 6462
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8459424686163207e-06,
+      "loss": 2.4568,
+      "step": 6463
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.843594997338647e-06,
+      "loss": 2.4564,
+      "step": 6464
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8412488680707464e-06,
+      "loss": 2.4331,
+      "step": 6465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8389040811986292e-06,
+      "loss": 2.4459,
+      "step": 6466
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8365606371081057e-06,
+      "loss": 2.4329,
+      "step": 6467
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8342185361847443e-06,
+      "loss": 2.4046,
+      "step": 6468
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.831877778813903e-06,
+      "loss": 2.4456,
+      "step": 6469
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8295383653807186e-06,
+      "loss": 2.4031,
+      "step": 6470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8272002962701051e-06,
+      "loss": 2.4435,
+      "step": 6471
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.824863571866755e-06,
+      "loss": 2.4991,
+      "step": 6472
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.822528192555142e-06,
+      "loss": 2.4828,
+      "step": 6473
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8201941587195104e-06,
+      "loss": 2.4295,
+      "step": 6474
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8178614707438957e-06,
+      "loss": 2.4298,
+      "step": 6475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8155301290121053e-06,
+      "loss": 2.4241,
+      "step": 6476
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.813200133907721e-06,
+      "loss": 2.4287,
+      "step": 6477
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8108714858141097e-06,
+      "loss": 2.5501,
+      "step": 6478
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8085441851144136e-06,
+      "loss": 2.4578,
+      "step": 6479
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8062182321915544e-06,
+      "loss": 2.4667,
+      "step": 6480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8038936274282316e-06,
+      "loss": 2.4195,
+      "step": 6481
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8015703712069232e-06,
+      "loss": 2.4196,
+      "step": 6482
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7992484639098839e-06,
+      "loss": 2.3977,
+      "step": 6483
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7969279059191512e-06,
+      "loss": 2.4282,
+      "step": 6484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.79460869761653e-06,
+      "loss": 2.4443,
+      "step": 6485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7922908393836148e-06,
+      "loss": 2.4409,
+      "step": 6486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7899743316017716e-06,
+      "loss": 2.4641,
+      "step": 6487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7876591746521465e-06,
+      "loss": 2.3747,
+      "step": 6488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.785345368915662e-06,
+      "loss": 2.4896,
+      "step": 6489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7830329147730218e-06,
+      "loss": 2.4733,
+      "step": 6490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7807218126046965e-06,
+      "loss": 2.4322,
+      "step": 6491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7784120627909506e-06,
+      "loss": 2.5048,
+      "step": 6492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.776103665711817e-06,
+      "loss": 2.5207,
+      "step": 6493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7737966217471004e-06,
+      "loss": 2.4271,
+      "step": 6494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7714909312763906e-06,
+      "loss": 2.5058,
+      "step": 6495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7691865946790598e-06,
+      "loss": 2.4357,
+      "step": 6496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7668836123342447e-06,
+      "loss": 2.4575,
+      "step": 6497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7645819846208655e-06,
+      "loss": 2.501,
+      "step": 6498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7622817119176205e-06,
+      "loss": 2.4272,
+      "step": 6499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7599827946029847e-06,
+      "loss": 2.4214,
+      "step": 6500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7576852330552107e-06,
+      "loss": 2.4524,
+      "step": 6501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.755389027652321e-06,
+      "loss": 2.5216,
+      "step": 6502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7530941787721246e-06,
+      "loss": 2.5767,
+      "step": 6503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7508006867922024e-06,
+      "loss": 2.475,
+      "step": 6504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7485085520899125e-06,
+      "loss": 2.341,
+      "step": 6505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7462177750423903e-06,
+      "loss": 2.4893,
+      "step": 6506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7439283560265508e-06,
+      "loss": 2.4631,
+      "step": 6507
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7416402954190748e-06,
+      "loss": 2.4658,
+      "step": 6508
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7393535935964333e-06,
+      "loss": 2.3777,
+      "step": 6509
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.737068250934868e-06,
+      "loss": 2.3739,
+      "step": 6510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7347842678103932e-06,
+      "loss": 2.4529,
+      "step": 6511
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7325016445988008e-06,
+      "loss": 2.4911,
+      "step": 6512
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7302203816756692e-06,
+      "loss": 2.4396,
+      "step": 6513
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7279404794163379e-06,
+      "loss": 2.338,
+      "step": 6514
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7256619381959317e-06,
+      "loss": 2.4613,
+      "step": 6515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.723384758389348e-06,
+      "loss": 2.3986,
+      "step": 6516
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7211089403712634e-06,
+      "loss": 2.5099,
+      "step": 6517
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7188344845161286e-06,
+      "loss": 2.4682,
+      "step": 6518
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7165613911981672e-06,
+      "loss": 2.4483,
+      "step": 6519
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7142896607913795e-06,
+      "loss": 2.4575,
+      "step": 6520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7120192936695513e-06,
+      "loss": 2.5117,
+      "step": 6521
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70975029020623e-06,
+      "loss": 2.4584,
+      "step": 6522
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7074826507747477e-06,
+      "loss": 2.4615,
+      "step": 6523
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7052163757482088e-06,
+      "loss": 2.3422,
+      "step": 6524
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029514654994895e-06,
+      "loss": 2.4406,
+      "step": 6525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7006879204012517e-06,
+      "loss": 2.4855,
+      "step": 6526
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6984257408259252e-06,
+      "loss": 2.4726,
+      "step": 6527
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6961649271457137e-06,
+      "loss": 2.4994,
+      "step": 6528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6939054797325982e-06,
+      "loss": 2.4341,
+      "step": 6529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6916473989583426e-06,
+      "loss": 2.3986,
+      "step": 6530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6893906851944731e-06,
+      "loss": 2.4639,
+      "step": 6531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.687135338812298e-06,
+      "loss": 2.4641,
+      "step": 6532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6848813601828995e-06,
+      "loss": 2.4534,
+      "step": 6533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6826287496771355e-06,
+      "loss": 2.4662,
+      "step": 6534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6803775076656392e-06,
+      "loss": 2.5436,
+      "step": 6535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6781276345188146e-06,
+      "loss": 2.5983,
+      "step": 6536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6758791306068422e-06,
+      "loss": 2.4554,
+      "step": 6537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6736319962996862e-06,
+      "loss": 2.3961,
+      "step": 6538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6713862319670693e-06,
+      "loss": 2.486,
+      "step": 6539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6691418379785008e-06,
+      "loss": 2.5383,
+      "step": 6540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6668988147032606e-06,
+      "loss": 2.3978,
+      "step": 6541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6646571625104025e-06,
+      "loss": 2.5121,
+      "step": 6542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6624168817687568e-06,
+      "loss": 2.4843,
+      "step": 6543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.660177972846928e-06,
+      "loss": 2.5231,
+      "step": 6544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.65794043611329e-06,
+      "loss": 2.5504,
+      "step": 6545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.655704271935995e-06,
+      "loss": 2.3827,
+      "step": 6546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6534694806829753e-06,
+      "loss": 2.3595,
+      "step": 6547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6512360627219238e-06,
+      "loss": 2.451,
+      "step": 6548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6490040184203183e-06,
+      "loss": 2.5174,
+      "step": 6549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6467733481454063e-06,
+      "loss": 2.3868,
+      "step": 6550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6445440522642097e-06,
+      "loss": 2.4562,
+      "step": 6551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6423161311435243e-06,
+      "loss": 2.365,
+      "step": 6552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6400895851499231e-06,
+      "loss": 2.423,
+      "step": 6553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6378644146497436e-06,
+      "loss": 2.4472,
+      "step": 6554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.63564062000911e-06,
+      "loss": 2.4498,
+      "step": 6555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.633418201593907e-06,
+      "loss": 2.5064,
+      "step": 6556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6311971597698017e-06,
+      "loss": 2.4161,
+      "step": 6557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.628977494902233e-06,
+      "loss": 2.4897,
+      "step": 6558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6267592073564098e-06,
+      "loss": 2.4406,
+      "step": 6559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.624542297497319e-06,
+      "loss": 2.401,
+      "step": 6560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6223267656897213e-06,
+      "loss": 2.4246,
+      "step": 6561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6201126122981414e-06,
+      "loss": 2.2944,
+      "step": 6562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6178998376868848e-06,
+      "loss": 2.4557,
+      "step": 6563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.615688442220037e-06,
+      "loss": 2.4398,
+      "step": 6564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.61347842626144e-06,
+      "loss": 2.4084,
+      "step": 6565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.611269790174722e-06,
+      "loss": 2.5126,
+      "step": 6566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.609062534323279e-06,
+      "loss": 2.5495,
+      "step": 6567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6068566590702794e-06,
+      "loss": 2.4015,
+      "step": 6568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6046521647786672e-06,
+      "loss": 2.5953,
+      "step": 6569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.60244905181116e-06,
+      "loss": 2.3768,
+      "step": 6570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6002473205302382e-06,
+      "loss": 2.5173,
+      "step": 6571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5980469712981717e-06,
+      "loss": 2.37,
+      "step": 6572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5958480044769874e-06,
+      "loss": 2.4399,
+      "step": 6573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.593650420428493e-06,
+      "loss": 2.4682,
+      "step": 6574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5914542195142667e-06,
+      "loss": 2.4324,
+      "step": 6575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5892594020956608e-06,
+      "loss": 2.4828,
+      "step": 6576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5870659685337964e-06,
+      "loss": 2.4703,
+      "step": 6577
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5848739191895723e-06,
+      "loss": 2.4315,
+      "step": 6578
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5826832544236481e-06,
+      "loss": 2.3619,
+      "step": 6579
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.580493974596472e-06,
+      "loss": 2.4081,
+      "step": 6580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5783060800682558e-06,
+      "loss": 2.5454,
+      "step": 6581
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5761195711989786e-06,
+      "loss": 2.4441,
+      "step": 6582
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5739344483483987e-06,
+      "loss": 2.4614,
+      "step": 6583
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5717507118760444e-06,
+      "loss": 2.3788,
+      "step": 6584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.569568362141215e-06,
+      "loss": 2.4238,
+      "step": 6585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5673873995029831e-06,
+      "loss": 2.4818,
+      "step": 6586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.565207824320194e-06,
+      "loss": 2.5218,
+      "step": 6587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5630296369514563e-06,
+      "loss": 2.3765,
+      "step": 6588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5608528377551657e-06,
+      "loss": 2.4624,
+      "step": 6589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5586774270894744e-06,
+      "loss": 2.5647,
+      "step": 6590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.556503405312314e-06,
+      "loss": 2.5709,
+      "step": 6591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5543307727813872e-06,
+      "loss": 2.4321,
+      "step": 6592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5521595298541658e-06,
+      "loss": 2.4879,
+      "step": 6593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.549989676887894e-06,
+      "loss": 2.3859,
+      "step": 6594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.547821214239591e-06,
+      "loss": 2.4094,
+      "step": 6595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5456541422660354e-06,
+      "loss": 2.4866,
+      "step": 6596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5434884613237933e-06,
+      "loss": 2.3678,
+      "step": 6597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5413241717691928e-06,
+      "loss": 2.4393,
+      "step": 6598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5391612739583295e-06,
+      "loss": 2.4755,
+      "step": 6599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5369997682470783e-06,
+      "loss": 2.4134,
+      "step": 6600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5348396549910804e-06,
+      "loss": 2.4365,
+      "step": 6601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5326809345457493e-06,
+      "loss": 2.4918,
+      "step": 6602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5305236072662677e-06,
+      "loss": 2.4777,
+      "step": 6603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5283676735075936e-06,
+      "loss": 2.415,
+      "step": 6604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5262131336244469e-06,
+      "loss": 2.4279,
+      "step": 6605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5240599879713303e-06,
+      "loss": 2.5041,
+      "step": 6606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5219082369025051e-06,
+      "loss": 2.5184,
+      "step": 6607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5197578807720093e-06,
+      "loss": 2.4315,
+      "step": 6608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.517608919933653e-06,
+      "loss": 2.408,
+      "step": 6609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5154613547410125e-06,
+      "loss": 2.4375,
+      "step": 6610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.513315185547437e-06,
+      "loss": 2.4543,
+      "step": 6611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5111704127060478e-06,
+      "loss": 2.4553,
+      "step": 6612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5090270365697268e-06,
+      "loss": 2.4463,
+      "step": 6613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5068850574911398e-06,
+      "loss": 2.5079,
+      "step": 6614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.504744475822716e-06,
+      "loss": 2.4119,
+      "step": 6615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5026052919166513e-06,
+      "loss": 2.5105,
+      "step": 6616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5004675061249174e-06,
+      "loss": 2.4272,
+      "step": 6617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4983311187992522e-06,
+      "loss": 2.5041,
+      "step": 6618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4961961302911664e-06,
+      "loss": 2.5386,
+      "step": 6619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4940625409519393e-06,
+      "loss": 2.435,
+      "step": 6620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4919303511326222e-06,
+      "loss": 2.5127,
+      "step": 6621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4897995611840265e-06,
+      "loss": 2.3979,
+      "step": 6622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4876701714567499e-06,
+      "loss": 2.5056,
+      "step": 6623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4855421823011428e-06,
+      "loss": 2.4591,
+      "step": 6624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4834155940673356e-06,
+      "loss": 2.4855,
+      "step": 6625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.481290407105227e-06,
+      "loss": 2.44,
+      "step": 6626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4791666217644806e-06,
+      "loss": 2.3941,
+      "step": 6627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4770442383945337e-06,
+      "loss": 2.3708,
+      "step": 6628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4749232573445949e-06,
+      "loss": 2.4318,
+      "step": 6629
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4728036789636301e-06,
+      "loss": 2.4475,
+      "step": 6630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4706855036003896e-06,
+      "loss": 2.475,
+      "step": 6631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4685687316033881e-06,
+      "loss": 2.4164,
+      "step": 6632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4664533633209022e-06,
+      "loss": 2.4427,
+      "step": 6633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4643393991009847e-06,
+      "loss": 2.4012,
+      "step": 6634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4622268392914562e-06,
+      "loss": 2.4354,
+      "step": 6635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4601156842399055e-06,
+      "loss": 2.4491,
+      "step": 6636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4580059342936902e-06,
+      "loss": 2.3978,
+      "step": 6637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4558975897999362e-06,
+      "loss": 2.427,
+      "step": 6638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.45379065110554e-06,
+      "loss": 2.4361,
+      "step": 6639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.451685118557168e-06,
+      "loss": 2.4876,
+      "step": 6640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4495809925012471e-06,
+      "loss": 2.4722,
+      "step": 6641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.447478273283981e-06,
+      "loss": 2.3859,
+      "step": 6642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4453769612513413e-06,
+      "loss": 2.4496,
+      "step": 6643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4432770567490639e-06,
+      "loss": 2.4161,
+      "step": 6644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.441178560122657e-06,
+      "loss": 2.4877,
+      "step": 6645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4390814717173972e-06,
+      "loss": 2.5156,
+      "step": 6646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4369857918783215e-06,
+      "loss": 2.5379,
+      "step": 6647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.434891520950248e-06,
+      "loss": 2.4257,
+      "step": 6648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4327986592777565e-06,
+      "loss": 2.48,
+      "step": 6649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4307072072051898e-06,
+      "loss": 2.5045,
+      "step": 6650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4286171650766678e-06,
+      "loss": 2.405,
+      "step": 6651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4265285332360724e-06,
+      "loss": 2.5023,
+      "step": 6652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4244413120270572e-06,
+      "loss": 2.4138,
+      "step": 6653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4223555017930413e-06,
+      "loss": 2.4988,
+      "step": 6654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.420271102877212e-06,
+      "loss": 2.4371,
+      "step": 6655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4181881156225252e-06,
+      "loss": 2.5037,
+      "step": 6656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4161065403717057e-06,
+      "loss": 2.3979,
+      "step": 6657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4140263774672414e-06,
+      "loss": 2.5274,
+      "step": 6658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.411947627251392e-06,
+      "loss": 2.498,
+      "step": 6659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4098702900661832e-06,
+      "loss": 2.4214,
+      "step": 6660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.40779436625341e-06,
+      "loss": 2.4874,
+      "step": 6661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4057198561546316e-06,
+      "loss": 2.3266,
+      "step": 6662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4036467601111803e-06,
+      "loss": 2.5386,
+      "step": 6663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4015750784641436e-06,
+      "loss": 2.3465,
+      "step": 6664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.399504811554393e-06,
+      "loss": 2.5216,
+      "step": 6665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3974359597225573e-06,
+      "loss": 2.4798,
+      "step": 6666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3953685233090297e-06,
+      "loss": 2.4294,
+      "step": 6667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.393302502653977e-06,
+      "loss": 2.4431,
+      "step": 6668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3912378980973319e-06,
+      "loss": 2.4486,
+      "step": 6669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3891747099787921e-06,
+      "loss": 2.3402,
+      "step": 6670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3871129386378234e-06,
+      "loss": 2.5433,
+      "step": 6671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3850525844136575e-06,
+      "loss": 2.4391,
+      "step": 6672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3829936476452954e-06,
+      "loss": 2.404,
+      "step": 6673
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.380936128671503e-06,
+      "loss": 2.4227,
+      "step": 6674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3788800278308111e-06,
+      "loss": 2.4642,
+      "step": 6675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3768253454615165e-06,
+      "loss": 2.4301,
+      "step": 6676
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3747720819016952e-06,
+      "loss": 2.3717,
+      "step": 6677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3727202374891702e-06,
+      "loss": 2.4273,
+      "step": 6678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3706698125615448e-06,
+      "loss": 2.4651,
+      "step": 6679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3686208074561858e-06,
+      "loss": 2.5105,
+      "step": 6680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.366573222510219e-06,
+      "loss": 2.4486,
+      "step": 6681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3645270580605486e-06,
+      "loss": 2.4557,
+      "step": 6682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3624823144438405e-06,
+      "loss": 2.4001,
+      "step": 6683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3604389919965199e-06,
+      "loss": 2.4533,
+      "step": 6684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3583970910547862e-06,
+      "loss": 2.3877,
+      "step": 6685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3563566119546024e-06,
+      "loss": 2.4061,
+      "step": 6686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3543175550316978e-06,
+      "loss": 2.4603,
+      "step": 6687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3522799206215686e-06,
+      "loss": 2.4801,
+      "step": 6688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3502437090594745e-06,
+      "loss": 2.5631,
+      "step": 6689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3482089206804428e-06,
+      "loss": 2.4572,
+      "step": 6690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.346175555819269e-06,
+      "loss": 2.553,
+      "step": 6691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.344143614810507e-06,
+      "loss": 2.4962,
+      "step": 6692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.342113097988481e-06,
+      "loss": 2.4831,
+      "step": 6693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3400840056872876e-06,
+      "loss": 2.3914,
+      "step": 6694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3380563382407763e-06,
+      "loss": 2.3579,
+      "step": 6695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.33603009598257e-06,
+      "loss": 2.5024,
+      "step": 6696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3340052792460557e-06,
+      "loss": 2.3797,
+      "step": 6697
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3319818883643854e-06,
+      "loss": 2.4873,
+      "step": 6698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3299599236704763e-06,
+      "loss": 2.3176,
+      "step": 6699
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3279393854970136e-06,
+      "loss": 2.4235,
+      "step": 6700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.325920274176441e-06,
+      "loss": 2.4633,
+      "step": 6701
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.323902590040973e-06,
+      "loss": 2.5077,
+      "step": 6702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.321886333422594e-06,
+      "loss": 2.4554,
+      "step": 6703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3198715046530408e-06,
+      "loss": 2.5019,
+      "step": 6704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.317858104063825e-06,
+      "loss": 2.5404,
+      "step": 6705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3158461319862203e-06,
+      "loss": 2.4216,
+      "step": 6706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3138355887512644e-06,
+      "loss": 2.4583,
+      "step": 6707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3118264746897647e-06,
+      "loss": 2.4833,
+      "step": 6708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3098187901322844e-06,
+      "loss": 2.5439,
+      "step": 6709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3078125354091563e-06,
+      "loss": 2.5416,
+      "step": 6710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3058077108504874e-06,
+      "loss": 2.4258,
+      "step": 6711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3038043167861314e-06,
+      "loss": 2.4164,
+      "step": 6712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.301802353545718e-06,
+      "loss": 2.5173,
+      "step": 6713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2998018214586416e-06,
+      "loss": 2.4986,
+      "step": 6714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2978027208540555e-06,
+      "loss": 2.4251,
+      "step": 6715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2958050520608822e-06,
+      "loss": 2.4603,
+      "step": 6716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2938088154078098e-06,
+      "loss": 2.4155,
+      "step": 6717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2918140112232836e-06,
+      "loss": 2.3425,
+      "step": 6718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2898206398355162e-06,
+      "loss": 2.4594,
+      "step": 6719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2878287015724933e-06,
+      "loss": 2.382,
+      "step": 6720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2858381967619516e-06,
+      "loss": 2.4319,
+      "step": 6721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2838491257313978e-06,
+      "loss": 2.3459,
+      "step": 6722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.281861488808106e-06,
+      "loss": 2.4855,
+      "step": 6723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2798752863191077e-06,
+      "loss": 2.488,
+      "step": 6724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2778905185912061e-06,
+      "loss": 2.4453,
+      "step": 6725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2759071859509576e-06,
+      "loss": 2.475,
+      "step": 6726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2739252887246911e-06,
+      "loss": 2.39,
+      "step": 6727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2719448272385015e-06,
+      "loss": 2.5322,
+      "step": 6728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2699658018182381e-06,
+      "loss": 2.4711,
+      "step": 6729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2679882127895194e-06,
+      "loss": 2.3766,
+      "step": 6730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2660120604777272e-06,
+      "loss": 2.5112,
+      "step": 6731
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.264037345208008e-06,
+      "loss": 2.4868,
+      "step": 6732
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2620640673052698e-06,
+      "loss": 2.4574,
+      "step": 6733
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2600922270941861e-06,
+      "loss": 2.4945,
+      "step": 6734
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2581218248991878e-06,
+      "loss": 2.5073,
+      "step": 6735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2561528610444795e-06,
+      "loss": 2.4147,
+      "step": 6736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2541853358540236e-06,
+      "loss": 2.4181,
+      "step": 6737
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2522192496515406e-06,
+      "loss": 2.3616,
+      "step": 6738
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2502546027605233e-06,
+      "loss": 2.4297,
+      "step": 6739
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.248291395504223e-06,
+      "loss": 2.4758,
+      "step": 6740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.246329628205656e-06,
+      "loss": 2.2704,
+      "step": 6741
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2443693011876e-06,
+      "loss": 2.4446,
+      "step": 6742
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2424104147725957e-06,
+      "loss": 2.3568,
+      "step": 6743
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2404529692829448e-06,
+      "loss": 2.4912,
+      "step": 6744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.238496965040722e-06,
+      "loss": 2.4739,
+      "step": 6745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2365424023677508e-06,
+      "loss": 2.4756,
+      "step": 6746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2345892815856265e-06,
+      "loss": 2.3768,
+      "step": 6747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.232637603015705e-06,
+      "loss": 2.415,
+      "step": 6748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2306873669791043e-06,
+      "loss": 2.4267,
+      "step": 6749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.228738573796704e-06,
+      "loss": 2.5504,
+      "step": 6750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2267912237891534e-06,
+      "loss": 2.4494,
+      "step": 6751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2248453172768481e-06,
+      "loss": 2.4514,
+      "step": 6752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.222900854579967e-06,
+      "loss": 2.4921,
+      "step": 6753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2209578360184393e-06,
+      "loss": 2.5411,
+      "step": 6754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.219016261911954e-06,
+      "loss": 2.4801,
+      "step": 6755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2170761325799706e-06,
+      "loss": 2.4134,
+      "step": 6756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2151374483417055e-06,
+      "loss": 2.4792,
+      "step": 6757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2132002095161398e-06,
+      "loss": 2.3151,
+      "step": 6758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2112644164220166e-06,
+      "loss": 2.4875,
+      "step": 6759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.209330069377842e-06,
+      "loss": 2.479,
+      "step": 6760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.207397168701877e-06,
+      "loss": 2.4925,
+      "step": 6761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2054657147121585e-06,
+      "loss": 2.4237,
+      "step": 6762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.203535707726471e-06,
+      "loss": 2.4066,
+      "step": 6763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.201607148062368e-06,
+      "loss": 2.4404,
+      "step": 6764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1996800360371675e-06,
+      "loss": 2.4582,
+      "step": 6765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1977543719679418e-06,
+      "loss": 2.4308,
+      "step": 6766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1958301561715313e-06,
+      "loss": 2.537,
+      "step": 6767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.193907388964538e-06,
+      "loss": 2.3975,
+      "step": 6768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1919860706633158e-06,
+      "loss": 2.3719,
+      "step": 6769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1900662015839949e-06,
+      "loss": 2.4691,
+      "step": 6770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1881477820424603e-06,
+      "loss": 2.4784,
+      "step": 6771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1862308123543532e-06,
+      "loss": 2.4704,
+      "step": 6772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1843152928350843e-06,
+      "loss": 2.3765,
+      "step": 6773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1824012237998206e-06,
+      "loss": 2.5285,
+      "step": 6774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1804886055634946e-06,
+      "loss": 2.5287,
+      "step": 6775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1785774384407967e-06,
+      "loss": 2.4412,
+      "step": 6776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.176667722746182e-06,
+      "loss": 2.4293,
+      "step": 6777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1747594587938592e-06,
+      "loss": 2.4203,
+      "step": 6778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1728526468978107e-06,
+      "loss": 2.4571,
+      "step": 6779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1709472873717676e-06,
+      "loss": 2.4693,
+      "step": 6780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1690433805292278e-06,
+      "loss": 2.4644,
+      "step": 6781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.167140926683451e-06,
+      "loss": 2.5314,
+      "step": 6782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1652399261474556e-06,
+      "loss": 2.441,
+      "step": 6783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1633403792340225e-06,
+      "loss": 2.4763,
+      "step": 6784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.161442286255694e-06,
+      "loss": 2.4336,
+      "step": 6785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1595456475247657e-06,
+      "loss": 2.4874,
+      "step": 6786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.157650463353307e-06,
+      "loss": 2.4485,
+      "step": 6787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1557567340531394e-06,
+      "loss": 2.4017,
+      "step": 6788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.153864459935844e-06,
+      "loss": 2.434,
+      "step": 6789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1519736413127668e-06,
+      "loss": 2.5513,
+      "step": 6790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1500842784950128e-06,
+      "loss": 2.5471,
+      "step": 6791
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1481963717934475e-06,
+      "loss": 2.4659,
+      "step": 6792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1463099215186957e-06,
+      "loss": 2.3739,
+      "step": 6793
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1444249279811438e-06,
+      "loss": 2.3342,
+      "step": 6794
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1425413914909377e-06,
+      "loss": 2.4233,
+      "step": 6795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1406593123579878e-06,
+      "loss": 2.4126,
+      "step": 6796
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1387786908919562e-06,
+      "loss": 2.4352,
+      "step": 6797
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.136899527402271e-06,
+      "loss": 2.4029,
+      "step": 6798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1350218221981203e-06,
+      "loss": 2.3474,
+      "step": 6799
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1331455755884512e-06,
+      "loss": 2.448,
+      "step": 6800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1312707878819706e-06,
+      "loss": 2.3946,
+      "step": 6801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1293974593871493e-06,
+      "loss": 2.4304,
+      "step": 6802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.127525590412205e-06,
+      "loss": 2.4197,
+      "step": 6803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.125655181265134e-06,
+      "loss": 2.5288,
+      "step": 6804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1237862322536819e-06,
+      "loss": 2.4324,
+      "step": 6805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1219187436853508e-06,
+      "loss": 2.4128,
+      "step": 6806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1200527158674102e-06,
+      "loss": 2.5504,
+      "step": 6807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1181881491068835e-06,
+      "loss": 2.5275,
+      "step": 6808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.116325043710559e-06,
+      "loss": 2.4109,
+      "step": 6809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1144633999849797e-06,
+      "loss": 2.5164,
+      "step": 6810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1126032182364522e-06,
+      "loss": 2.5074,
+      "step": 6811
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.110744498771038e-06,
+      "loss": 2.4627,
+      "step": 6812
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1088872418945651e-06,
+      "loss": 2.355,
+      "step": 6813
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1070314479126098e-06,
+      "loss": 2.3696,
+      "step": 6814
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1051771171305182e-06,
+      "loss": 2.4077,
+      "step": 6815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1033242498533913e-06,
+      "loss": 2.5184,
+      "step": 6816
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.101472846386088e-06,
+      "loss": 2.5175,
+      "step": 6817
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0996229070332297e-06,
+      "loss": 2.4208,
+      "step": 6818
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0977744320991968e-06,
+      "loss": 2.3336,
+      "step": 6819
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.095927421888121e-06,
+      "loss": 2.3304,
+      "step": 6820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0940818767039053e-06,
+      "loss": 2.5437,
+      "step": 6821
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.092237796850205e-06,
+      "loss": 2.4532,
+      "step": 6822
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0903951826304314e-06,
+      "loss": 2.3917,
+      "step": 6823
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0885540343477597e-06,
+      "loss": 2.4469,
+      "step": 6824
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0867143523051215e-06,
+      "loss": 2.5164,
+      "step": 6825
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0848761368052097e-06,
+      "loss": 2.4656,
+      "step": 6826
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.083039388150473e-06,
+      "loss": 2.4724,
+      "step": 6827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0812041066431201e-06,
+      "loss": 2.4739,
+      "step": 6828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0793702925851179e-06,
+      "loss": 2.4627,
+      "step": 6829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0775379462781943e-06,
+      "loss": 2.5074,
+      "step": 6830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.075707068023829e-06,
+      "loss": 2.4442,
+      "step": 6831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0738776581232658e-06,
+      "loss": 2.4751,
+      "step": 6832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.07204971687751e-06,
+      "loss": 2.4399,
+      "step": 6833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.070223244587315e-06,
+      "loss": 2.4657,
+      "step": 6834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0683982415532024e-06,
+      "loss": 2.3855,
+      "step": 6835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0665747080754474e-06,
+      "loss": 2.4701,
+      "step": 6836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0647526444540811e-06,
+      "loss": 2.4047,
+      "step": 6837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0629320509888996e-06,
+      "loss": 2.429,
+      "step": 6838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.061112927979453e-06,
+      "loss": 2.4298,
+      "step": 6839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0592952757250474e-06,
+      "loss": 2.4092,
+      "step": 6840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0574790945247505e-06,
+      "loss": 2.3828,
+      "step": 6841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0556643846773874e-06,
+      "loss": 2.558,
+      "step": 6842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0538511464815382e-06,
+      "loss": 2.4913,
+      "step": 6843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0520393802355454e-06,
+      "loss": 2.3502,
+      "step": 6844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0502290862375063e-06,
+      "loss": 2.4648,
+      "step": 6845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.048420264785276e-06,
+      "loss": 2.4145,
+      "step": 6846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0466129161764716e-06,
+      "loss": 2.47,
+      "step": 6847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0448070407084586e-06,
+      "loss": 2.4284,
+      "step": 6848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0430026386783665e-06,
+      "loss": 2.477,
+      "step": 6849
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0411997103830862e-06,
+      "loss": 2.5022,
+      "step": 6850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0393982561192584e-06,
+      "loss": 2.4444,
+      "step": 6851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0375982761832847e-06,
+      "loss": 2.4414,
+      "step": 6852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.035799770871323e-06,
+      "loss": 2.4711,
+      "step": 6853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0340027404792919e-06,
+      "loss": 2.4513,
+      "step": 6854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0322071853028615e-06,
+      "loss": 2.4495,
+      "step": 6855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0304131056374678e-06,
+      "loss": 2.4853,
+      "step": 6856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0286205017782923e-06,
+      "loss": 2.4034,
+      "step": 6857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.026829374020284e-06,
+      "loss": 2.3941,
+      "step": 6858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0250397226581444e-06,
+      "loss": 2.4942,
+      "step": 6859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0232515479863325e-06,
+      "loss": 2.3821,
+      "step": 6860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0214648502990643e-06,
+      "loss": 2.4305,
+      "step": 6861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0196796298903145e-06,
+      "loss": 2.4791,
+      "step": 6862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0178958870538125e-06,
+      "loss": 2.5021,
+      "step": 6863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0161136220830469e-06,
+      "loss": 2.446,
+      "step": 6864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0143328352712589e-06,
+      "loss": 2.4763,
+      "step": 6865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0125535269114494e-06,
+      "loss": 2.5807,
+      "step": 6866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0107756972963812e-06,
+      "loss": 2.5885,
+      "step": 6867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0089993467185621e-06,
+      "loss": 2.4504,
+      "step": 6868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0072244754702666e-06,
+      "loss": 2.517,
+      "step": 6869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0054510838435206e-06,
+      "loss": 2.6066,
+      "step": 6870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0036791721301088e-06,
+      "loss": 2.3924,
+      "step": 6871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.001908740621572e-06,
+      "loss": 2.4978,
+      "step": 6872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0001397896092092e-06,
+      "loss": 2.3668,
+      "step": 6873
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.9837231938407e-07,
+      "loss": 2.4564,
+      "step": 6874
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.966063302369655e-07,
+      "loss": 2.4915,
+      "step": 6875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.948418224584611e-07,
+      "loss": 2.4582,
+      "step": 6876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.93078796338881e-07,
+      "loss": 2.4345,
+      "step": 6877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.913172521683023e-07,
+      "loss": 2.5211,
+      "step": 6878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.895571902365608e-07,
+      "loss": 2.5276,
+      "step": 6879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.877986108332472e-07,
+      "loss": 2.3298,
+      "step": 6880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.860415142477097e-07,
+      "loss": 2.4104,
+      "step": 6881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.842859007690476e-07,
+      "loss": 2.4402,
+      "step": 6882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.825317706861204e-07,
+      "loss": 2.4385,
+      "step": 6883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.807791242875474e-07,
+      "loss": 2.3401,
+      "step": 6884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.790279618616961e-07,
+      "loss": 2.3788,
+      "step": 6885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.77278283696692e-07,
+      "loss": 2.5288,
+      "step": 6886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.755300900804187e-07,
+      "loss": 2.3602,
+      "step": 6887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.73783381300516e-07,
+      "loss": 2.424,
+      "step": 6888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.72038157644375e-07,
+      "loss": 2.4765,
+      "step": 6889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.702944193991482e-07,
+      "loss": 2.4893,
+      "step": 6890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.68552166851735e-07,
+      "loss": 2.4493,
+      "step": 6891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.66811400288804e-07,
+      "loss": 2.4682,
+      "step": 6892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.65072119996765e-07,
+      "loss": 2.4293,
+      "step": 6893
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.633343262617912e-07,
+      "loss": 2.424,
+      "step": 6894
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.615980193698105e-07,
+      "loss": 2.4481,
+      "step": 6895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.598631996065045e-07,
+      "loss": 2.4242,
+      "step": 6896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.581298672573115e-07,
+      "loss": 2.4984,
+      "step": 6897
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.563980226074244e-07,
+      "loss": 2.4336,
+      "step": 6898
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.546676659417908e-07,
+      "loss": 2.5283,
+      "step": 6899
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.52938797545111e-07,
+      "loss": 2.4865,
+      "step": 6900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.512114177018516e-07,
+      "loss": 2.5042,
+      "step": 6901
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.494855266962178e-07,
+      "loss": 2.418,
+      "step": 6902
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.477611248121832e-07,
+      "loss": 2.4666,
+      "step": 6903
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.460382123334689e-07,
+      "loss": 2.5003,
+      "step": 6904
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.443167895435535e-07,
+      "loss": 2.4064,
+      "step": 6905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.425968567256716e-07,
+      "loss": 2.4909,
+      "step": 6906
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.408784141628124e-07,
+      "loss": 2.4065,
+      "step": 6907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.391614621377132e-07,
+      "loss": 2.3373,
+      "step": 6908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.37446000932879e-07,
+      "loss": 2.4622,
+      "step": 6909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.357320308305606e-07,
+      "loss": 2.4043,
+      "step": 6910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.340195521127615e-07,
+      "loss": 2.4268,
+      "step": 6911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.323085650612462e-07,
+      "loss": 2.4347,
+      "step": 6912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.305990699575307e-07,
+      "loss": 2.4607,
+      "step": 6913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.288910670828865e-07,
+      "loss": 2.4844,
+      "step": 6914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.271845567183391e-07,
+      "loss": 2.4433,
+      "step": 6915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.254795391446658e-07,
+      "loss": 2.4516,
+      "step": 6916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.237760146424013e-07,
+      "loss": 2.4039,
+      "step": 6917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.220739834918379e-07,
+      "loss": 2.4951,
+      "step": 6918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.20373445973014e-07,
+      "loss": 2.3833,
+      "step": 6919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.18674402365729e-07,
+      "loss": 2.4641,
+      "step": 6920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.169768529495327e-07,
+      "loss": 2.4203,
+      "step": 6921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.152807980037315e-07,
+      "loss": 2.5123,
+      "step": 6922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.135862378073856e-07,
+      "loss": 2.3945,
+      "step": 6923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.118931726393087e-07,
+      "loss": 2.4284,
+      "step": 6924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.102016027780636e-07,
+      "loss": 2.4424,
+      "step": 6925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.085115285019774e-07,
+      "loss": 2.4923,
+      "step": 6926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.068229500891257e-07,
+      "loss": 2.4887,
+      "step": 6927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.051358678173339e-07,
+      "loss": 2.4646,
+      "step": 6928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.034502819641866e-07,
+      "loss": 2.4342,
+      "step": 6929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.01766192807022e-07,
+      "loss": 2.4163,
+      "step": 6930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.000836006229308e-07,
+      "loss": 2.5309,
+      "step": 6931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.98402505688758e-07,
+      "loss": 2.4912,
+      "step": 6932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.96722908281098e-07,
+      "loss": 2.4264,
+      "step": 6933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.950448086763042e-07,
+      "loss": 2.4349,
+      "step": 6934
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.933682071504856e-07,
+      "loss": 2.4463,
+      "step": 6935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.916931039794962e-07,
+      "loss": 2.4829,
+      "step": 6936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.900194994389499e-07,
+      "loss": 2.4453,
+      "step": 6937
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.883473938042119e-07,
+      "loss": 2.5143,
+      "step": 6938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.866767873504023e-07,
+      "loss": 2.4897,
+      "step": 6939
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.850076803523922e-07,
+      "loss": 2.4122,
+      "step": 6940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.833400730848097e-07,
+      "loss": 2.4629,
+      "step": 6941
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.816739658220286e-07,
+      "loss": 2.4512,
+      "step": 6942
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.800093588381864e-07,
+      "loss": 2.5735,
+      "step": 6943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.783462524071673e-07,
+      "loss": 2.5488,
+      "step": 6944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.766846468026058e-07,
+      "loss": 2.4553,
+      "step": 6945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.750245422978975e-07,
+      "loss": 2.4241,
+      "step": 6946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.733659391661841e-07,
+      "loss": 2.3144,
+      "step": 6947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.717088376803651e-07,
+      "loss": 2.499,
+      "step": 6948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.700532381130889e-07,
+      "loss": 2.4756,
+      "step": 6949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.683991407367598e-07,
+      "loss": 2.5465,
+      "step": 6950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.667465458235346e-07,
+      "loss": 2.4604,
+      "step": 6951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.650954536453226e-07,
+      "loss": 2.4462,
+      "step": 6952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.63445864473782e-07,
+      "loss": 2.3778,
+      "step": 6953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.617977785803289e-07,
+      "loss": 2.4227,
+      "step": 6954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.601511962361297e-07,
+      "loss": 2.4648,
+      "step": 6955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.585061177121057e-07,
+      "loss": 2.3952,
+      "step": 6956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.568625432789279e-07,
+      "loss": 2.4469,
+      "step": 6957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.552204732070224e-07,
+      "loss": 2.4299,
+      "step": 6958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.535799077665619e-07,
+      "loss": 2.4073,
+      "step": 6959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.519408472274804e-07,
+      "loss": 2.4657,
+      "step": 6960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.5030329185946e-07,
+      "loss": 2.4977,
+      "step": 6961
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.486672419319331e-07,
+      "loss": 2.4428,
+      "step": 6962
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.470326977140852e-07,
+      "loss": 2.4958,
+      "step": 6963
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.453996594748571e-07,
+      "loss": 2.444,
+      "step": 6964
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.43768127482939e-07,
+      "loss": 2.5011,
+      "step": 6965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.421381020067754e-07,
+      "loss": 2.4328,
+      "step": 6966
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.405095833145605e-07,
+      "loss": 2.5069,
+      "step": 6967
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.388825716742422e-07,
+      "loss": 2.5083,
+      "step": 6968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.372570673535218e-07,
+      "loss": 2.5039,
+      "step": 6969
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.356330706198467e-07,
+      "loss": 2.4496,
+      "step": 6970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.34010581740422e-07,
+      "loss": 2.3938,
+      "step": 6971
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.323896009822053e-07,
+      "loss": 2.4325,
+      "step": 6972
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.307701286119007e-07,
+      "loss": 2.4236,
+      "step": 6973
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.291521648959688e-07,
+      "loss": 2.4509,
+      "step": 6974
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.275357101006221e-07,
+      "loss": 2.4175,
+      "step": 6975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.259207644918188e-07,
+      "loss": 2.4833,
+      "step": 6976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.243073283352765e-07,
+      "loss": 2.4966,
+      "step": 6977
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.226954018964617e-07,
+      "loss": 2.4273,
+      "step": 6978
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.210849854405878e-07,
+      "loss": 2.5016,
+      "step": 6979
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.194760792326273e-07,
+      "loss": 2.4258,
+      "step": 6980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.178686835372995e-07,
+      "loss": 2.4254,
+      "step": 6981
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.162627986190774e-07,
+      "loss": 2.3685,
+      "step": 6982
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.146584247421829e-07,
+      "loss": 2.5147,
+      "step": 6983
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.130555621705915e-07,
+      "loss": 2.4419,
+      "step": 6984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.114542111680302e-07,
+      "loss": 2.5095,
+      "step": 6985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.098543719979768e-07,
+      "loss": 2.3859,
+      "step": 6986
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.082560449236576e-07,
+      "loss": 2.393,
+      "step": 6987
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.066592302080512e-07,
+      "loss": 2.4712,
+      "step": 6988
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.05063928113895e-07,
+      "loss": 2.434,
+      "step": 6989
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.034701389036659e-07,
+      "loss": 2.4462,
+      "step": 6990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.018778628395985e-07,
+      "loss": 2.3919,
+      "step": 6991
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.002871001836788e-07,
+      "loss": 2.3868,
+      "step": 6992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.986978511976374e-07,
+      "loss": 2.4588,
+      "step": 6993
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.971101161429651e-07,
+      "loss": 2.4206,
+      "step": 6994
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.955238952808996e-07,
+      "loss": 2.506,
+      "step": 6995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.939391888724246e-07,
+      "loss": 2.4398,
+      "step": 6996
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.923559971782824e-07,
+      "loss": 2.3642,
+      "step": 6997
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.907743204589602e-07,
+      "loss": 2.4064,
+      "step": 6998
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.891941589746999e-07,
+      "loss": 2.4962,
+      "step": 6999
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.876155129854934e-07,
+      "loss": 2.4609,
+      "step": 7000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.860383827510798e-07,
+      "loss": 2.506,
+      "step": 7001
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.844627685309536e-07,
+      "loss": 2.4365,
+      "step": 7002
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.828886705843585e-07,
+      "loss": 2.3871,
+      "step": 7003
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.813160891702842e-07,
+      "loss": 2.4647,
+      "step": 7004
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.797450245474746e-07,
+      "loss": 2.4855,
+      "step": 7005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.781754769744299e-07,
+      "loss": 2.4379,
+      "step": 7006
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.766074467093887e-07,
+      "loss": 2.4282,
+      "step": 7007
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.750409340103493e-07,
+      "loss": 2.4294,
+      "step": 7008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.734759391350544e-07,
+      "loss": 2.4646,
+      "step": 7009
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.719124623410024e-07,
+      "loss": 2.4557,
+      "step": 7010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.703505038854365e-07,
+      "loss": 2.4408,
+      "step": 7011
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.687900640253565e-07,
+      "loss": 2.373,
+      "step": 7012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.67231143017505e-07,
+      "loss": 2.4298,
+      "step": 7013
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.656737411183779e-07,
+      "loss": 2.4863,
+      "step": 7014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.641178585842235e-07,
+      "loss": 2.4778,
+      "step": 7015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.625634956710382e-07,
+      "loss": 2.4093,
+      "step": 7016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.610106526345673e-07,
+      "loss": 2.4601,
+      "step": 7017
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.594593297303066e-07,
+      "loss": 2.4455,
+      "step": 7018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.579095272135028e-07,
+      "loss": 2.4319,
+      "step": 7019
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.563612453391544e-07,
+      "loss": 2.4552,
+      "step": 7020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.548144843620031e-07,
+      "loss": 2.4463,
+      "step": 7021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.532692445365441e-07,
+      "loss": 2.5052,
+      "step": 7022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.517255261170275e-07,
+      "loss": 2.4573,
+      "step": 7023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.501833293574435e-07,
+      "loss": 2.4665,
+      "step": 7024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.486426545115378e-07,
+      "loss": 2.4446,
+      "step": 7025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.471035018328054e-07,
+      "loss": 2.4508,
+      "step": 7026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.455658715744906e-07,
+      "loss": 2.4901,
+      "step": 7027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.440297639895844e-07,
+      "loss": 2.4319,
+      "step": 7028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.424951793308322e-07,
+      "loss": 2.5421,
+      "step": 7029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.409621178507242e-07,
+      "loss": 2.4179,
+      "step": 7030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.394305798015011e-07,
+      "loss": 2.4824,
+      "step": 7031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.379005654351545e-07,
+      "loss": 2.3721,
+      "step": 7032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.363720750034254e-07,
+      "loss": 2.5572,
+      "step": 7033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.348451087578013e-07,
+      "loss": 2.3864,
+      "step": 7034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.333196669495235e-07,
+      "loss": 2.3663,
+      "step": 7035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.317957498295769e-07,
+      "loss": 2.4121,
+      "step": 7036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.30273357648702e-07,
+      "loss": 2.4778,
+      "step": 7037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.287524906573818e-07,
+      "loss": 2.5672,
+      "step": 7038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.272331491058493e-07,
+      "loss": 2.4626,
+      "step": 7039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.257153332440947e-07,
+      "loss": 2.4018,
+      "step": 7040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.24199043321846e-07,
+      "loss": 2.5138,
+      "step": 7041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.226842795885869e-07,
+      "loss": 2.4976,
+      "step": 7042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.21171042293548e-07,
+      "loss": 2.4246,
+      "step": 7043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.196593316857092e-07,
+      "loss": 2.5104,
+      "step": 7044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.181491480137992e-07,
+      "loss": 2.4514,
+      "step": 7045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.16640491526297e-07,
+      "loss": 2.4045,
+      "step": 7046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.151333624714229e-07,
+      "loss": 2.4682,
+      "step": 7047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.136277610971588e-07,
+      "loss": 2.4232,
+      "step": 7048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.121236876512227e-07,
+      "loss": 2.3516,
+      "step": 7049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.106211423810882e-07,
+      "loss": 2.3805,
+      "step": 7050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.091201255339763e-07,
+      "loss": 2.4167,
+      "step": 7051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.07620637356855e-07,
+      "loss": 2.4268,
+      "step": 7052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.061226780964425e-07,
+      "loss": 2.4757,
+      "step": 7053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.046262479992072e-07,
+      "loss": 2.4115,
+      "step": 7054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.031313473113588e-07,
+      "loss": 2.4378,
+      "step": 7055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.016379762788605e-07,
+      "loss": 2.4338,
+      "step": 7056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.001461351474281e-07,
+      "loss": 2.4538,
+      "step": 7057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.986558241625152e-07,
+      "loss": 2.4574,
+      "step": 7058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.971670435693334e-07,
+      "loss": 2.465,
+      "step": 7059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.956797936128357e-07,
+      "loss": 2.4083,
+      "step": 7060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.941940745377274e-07,
+      "loss": 2.4674,
+      "step": 7061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.927098865884607e-07,
+      "loss": 2.5537,
+      "step": 7062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.91227230009236e-07,
+      "loss": 2.4324,
+      "step": 7063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.89746105043998e-07,
+      "loss": 2.4419,
+      "step": 7064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.882665119364484e-07,
+      "loss": 2.4082,
+      "step": 7065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.867884509300271e-07,
+      "loss": 2.5349,
+      "step": 7066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.853119222679272e-07,
+      "loss": 2.4246,
+      "step": 7067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.838369261930878e-07,
+      "loss": 2.3838,
+      "step": 7068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.82363462948199e-07,
+      "loss": 2.456,
+      "step": 7069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.808915327756937e-07,
+      "loss": 2.4373,
+      "step": 7070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.794211359177571e-07,
+      "loss": 2.4102,
+      "step": 7071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.779522726163191e-07,
+      "loss": 2.4204,
+      "step": 7072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.76484943113056e-07,
+      "loss": 2.3833,
+      "step": 7073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.750191476493995e-07,
+      "loss": 2.372,
+      "step": 7074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.735548864665197e-07,
+      "loss": 2.4062,
+      "step": 7075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.720921598053387e-07,
+      "loss": 2.348,
+      "step": 7076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.70630967906526e-07,
+      "loss": 2.4662,
+      "step": 7077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.69171311010497e-07,
+      "loss": 2.5323,
+      "step": 7078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.677131893574162e-07,
+      "loss": 2.4273,
+      "step": 7079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.662566031871976e-07,
+      "loss": 2.5326,
+      "step": 7080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.648015527394925e-07,
+      "loss": 2.501,
+      "step": 7081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.633480382537161e-07,
+      "loss": 2.4897,
+      "step": 7082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.618960599690138e-07,
+      "loss": 2.5245,
+      "step": 7083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.604456181242913e-07,
+      "loss": 2.4463,
+      "step": 7084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.589967129581931e-07,
+      "loss": 2.4476,
+      "step": 7085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.575493447091152e-07,
+      "loss": 2.5156,
+      "step": 7086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.561035136151994e-07,
+      "loss": 2.381,
+      "step": 7087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.546592199143376e-07,
+      "loss": 2.5273,
+      "step": 7088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.532164638441607e-07,
+      "loss": 2.3962,
+      "step": 7089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.517752456420534e-07,
+      "loss": 2.4072,
+      "step": 7090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.503355655451493e-07,
+      "loss": 2.5011,
+      "step": 7091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.488974237903223e-07,
+      "loss": 2.4426,
+      "step": 7092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.474608206141963e-07,
+      "loss": 2.4745,
+      "step": 7093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.460257562531424e-07,
+      "loss": 2.4527,
+      "step": 7094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.445922309432784e-07,
+      "loss": 2.4216,
+      "step": 7095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.431602449204677e-07,
+      "loss": 2.4615,
+      "step": 7096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.417297984203252e-07,
+      "loss": 2.536,
+      "step": 7097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.403008916782017e-07,
+      "loss": 2.4758,
+      "step": 7098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.388735249292088e-07,
+      "loss": 2.4444,
+      "step": 7099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.374476984081923e-07,
+      "loss": 2.4945,
+      "step": 7100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.360234123497533e-07,
+      "loss": 2.3946,
+      "step": 7101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.346006669882332e-07,
+      "loss": 2.4031,
+      "step": 7102
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.331794625577248e-07,
+      "loss": 2.475,
+      "step": 7103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.317597992920632e-07,
+      "loss": 2.4176,
+      "step": 7104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.30341677424835e-07,
+      "loss": 2.5177,
+      "step": 7105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.289250971893657e-07,
+      "loss": 2.4757,
+      "step": 7106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.275100588187355e-07,
+      "loss": 2.4096,
+      "step": 7107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.260965625457671e-07,
+      "loss": 2.4897,
+      "step": 7108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.246846086030256e-07,
+      "loss": 2.3754,
+      "step": 7109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.232741972228295e-07,
+      "loss": 2.5075,
+      "step": 7110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.218653286372379e-07,
+      "loss": 2.4573,
+      "step": 7111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.204580030780593e-07,
+      "loss": 2.4516,
+      "step": 7112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.190522207768479e-07,
+      "loss": 2.4189,
+      "step": 7113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.176479819649039e-07,
+      "loss": 2.4308,
+      "step": 7114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.162452868732693e-07,
+      "loss": 2.4589,
+      "step": 7115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.148441357327384e-07,
+      "loss": 2.4201,
+      "step": 7116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.134445287738511e-07,
+      "loss": 2.5148,
+      "step": 7117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.120464662268876e-07,
+      "loss": 2.4248,
+      "step": 7118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.106499483218776e-07,
+      "loss": 2.4411,
+      "step": 7119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.092549752885979e-07,
+      "loss": 2.4818,
+      "step": 7120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.078615473565674e-07,
+      "loss": 2.3305,
+      "step": 7121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.064696647550572e-07,
+      "loss": 2.5114,
+      "step": 7122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.050793277130751e-07,
+      "loss": 2.5318,
+      "step": 7123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.036905364593826e-07,
+      "loss": 2.4971,
+      "step": 7124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.023032912224836e-07,
+      "loss": 2.4688,
+      "step": 7125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.009175922306265e-07,
+      "loss": 2.4562,
+      "step": 7126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.995334397118069e-07,
+      "loss": 2.4521,
+      "step": 7127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.981508338937647e-07,
+      "loss": 2.495,
+      "step": 7128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.96769775003988e-07,
+      "loss": 2.5408,
+      "step": 7129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.953902632697084e-07,
+      "loss": 2.4615,
+      "step": 7130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.940122989179043e-07,
+      "loss": 2.4576,
+      "step": 7131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.92635882175292e-07,
+      "loss": 2.3959,
+      "step": 7132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.912610132683472e-07,
+      "loss": 2.4915,
+      "step": 7133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.89887692423281e-07,
+      "loss": 2.5145,
+      "step": 7134
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.885159198660495e-07,
+      "loss": 2.3462,
+      "step": 7135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.871456958223587e-07,
+      "loss": 2.4462,
+      "step": 7136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.857770205176571e-07,
+      "loss": 2.3532,
+      "step": 7137
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.844098941771392e-07,
+      "loss": 2.5522,
+      "step": 7138
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.83044317025746e-07,
+      "loss": 2.3875,
+      "step": 7139
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.816802892881567e-07,
+      "loss": 2.4524,
+      "step": 7140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.803178111888064e-07,
+      "loss": 2.3813,
+      "step": 7141
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.789568829518688e-07,
+      "loss": 2.5248,
+      "step": 7142
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.775975048012617e-07,
+      "loss": 2.5219,
+      "step": 7143
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.762396769606482e-07,
+      "loss": 2.4693,
+      "step": 7144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.748833996534442e-07,
+      "loss": 2.4287,
+      "step": 7145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.735286731027979e-07,
+      "loss": 2.47,
+      "step": 7146
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.721754975316096e-07,
+      "loss": 2.4407,
+      "step": 7147
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.708238731625271e-07,
+      "loss": 2.3661,
+      "step": 7148
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.694738002179323e-07,
+      "loss": 2.4165,
+      "step": 7149
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.681252789199654e-07,
+      "loss": 2.3807,
+      "step": 7150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.667783094905032e-07,
+      "loss": 2.5053,
+      "step": 7151
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.654328921511653e-07,
+      "loss": 2.4557,
+      "step": 7152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.640890271233213e-07,
+      "loss": 2.4268,
+      "step": 7153
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.62746714628084e-07,
+      "loss": 2.3929,
+      "step": 7154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.614059548863094e-07,
+      "loss": 2.4124,
+      "step": 7155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.600667481185984e-07,
+      "loss": 2.4266,
+      "step": 7156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.587290945452983e-07,
+      "loss": 2.3826,
+      "step": 7157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.573929943864964e-07,
+      "loss": 2.4264,
+      "step": 7158
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.560584478620312e-07,
+      "loss": 2.4996,
+      "step": 7159
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.547254551914772e-07,
+      "loss": 2.4885,
+      "step": 7160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.533940165941576e-07,
+      "loss": 2.4282,
+      "step": 7161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.520641322891462e-07,
+      "loss": 2.4325,
+      "step": 7162
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.50735802495248e-07,
+      "loss": 2.5225,
+      "step": 7163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.494090274310227e-07,
+      "loss": 2.4902,
+      "step": 7164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.480838073147688e-07,
+      "loss": 2.4597,
+      "step": 7165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.467601423645308e-07,
+      "loss": 2.5674,
+      "step": 7166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.45438032798099e-07,
+      "loss": 2.4639,
+      "step": 7167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.44117478833005e-07,
+      "loss": 2.415,
+      "step": 7168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.427984806865238e-07,
+      "loss": 2.4734,
+      "step": 7169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.414810385756786e-07,
+      "loss": 2.4717,
+      "step": 7170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.401651527172325e-07,
+      "loss": 2.4681,
+      "step": 7171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.388508233276935e-07,
+      "loss": 2.5295,
+      "step": 7172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.375380506233163e-07,
+      "loss": 2.4217,
+      "step": 7173
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.36226834820095e-07,
+      "loss": 2.4751,
+      "step": 7174
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.349171761337713e-07,
+      "loss": 2.3306,
+      "step": 7175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.336090747798295e-07,
+      "loss": 2.3624,
+      "step": 7176
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.323025309734964e-07,
+      "loss": 2.4294,
+      "step": 7177
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.309975449297411e-07,
+      "loss": 2.4324,
+      "step": 7178
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.296941168632852e-07,
+      "loss": 2.4515,
+      "step": 7179
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.283922469885816e-07,
+      "loss": 2.3725,
+      "step": 7180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.270919355198345e-07,
+      "loss": 2.4786,
+      "step": 7181
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.257931826709917e-07,
+      "loss": 2.5561,
+      "step": 7182
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2449598865574e-07,
+      "loss": 2.4975,
+      "step": 7183
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.232003536875141e-07,
+      "loss": 2.5022,
+      "step": 7184
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.219062779794925e-07,
+      "loss": 2.4214,
+      "step": 7185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.206137617445917e-07,
+      "loss": 2.3911,
+      "step": 7186
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.193228051954769e-07,
+      "loss": 2.438,
+      "step": 7187
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.180334085445548e-07,
+      "loss": 2.3623,
+      "step": 7188
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.167455720039749e-07,
+      "loss": 2.4882,
+      "step": 7189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.154592957856319e-07,
+      "loss": 2.5069,
+      "step": 7190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.141745801011622e-07,
+      "loss": 2.4721,
+      "step": 7191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.128914251619444e-07,
+      "loss": 2.5366,
+      "step": 7192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.116098311791051e-07,
+      "loss": 2.4816,
+      "step": 7193
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.10329798363508e-07,
+      "loss": 2.5192,
+      "step": 7194
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.090513269257613e-07,
+      "loss": 2.5196,
+      "step": 7195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.077744170762211e-07,
+      "loss": 2.4272,
+      "step": 7196
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.064990690249816e-07,
+      "loss": 2.4349,
+      "step": 7197
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.052252829818805e-07,
+      "loss": 2.4425,
+      "step": 7198
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.039530591565011e-07,
+      "loss": 2.5232,
+      "step": 7199
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.026823977581674e-07,
+      "loss": 2.3875,
+      "step": 7200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.014132989959464e-07,
+      "loss": 2.4062,
+      "step": 7201
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.001457630786499e-07,
+      "loss": 2.4309,
+      "step": 7202
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.98879790214829e-07,
+      "loss": 2.3972,
+      "step": 7203
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.976153806127837e-07,
+      "loss": 2.457,
+      "step": 7204
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.963525344805497e-07,
+      "loss": 2.4754,
+      "step": 7205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.950912520259088e-07,
+      "loss": 2.4832,
+      "step": 7206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.93831533456387e-07,
+      "loss": 2.367,
+      "step": 7207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.925733789792508e-07,
+      "loss": 2.4759,
+      "step": 7208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.913167888015091e-07,
+      "loss": 2.4243,
+      "step": 7209
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.900617631299165e-07,
+      "loss": 2.4954,
+      "step": 7210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.888083021709666e-07,
+      "loss": 2.5044,
+      "step": 7211
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.875564061308946e-07,
+      "loss": 2.3986,
+      "step": 7212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.863060752156856e-07,
+      "loss": 2.4753,
+      "step": 7213
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.850573096310574e-07,
+      "loss": 2.446,
+      "step": 7214
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.838101095824776e-07,
+      "loss": 2.4089,
+      "step": 7215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.825644752751535e-07,
+      "loss": 2.3963,
+      "step": 7216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.813204069140343e-07,
+      "loss": 2.4935,
+      "step": 7217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.800779047038118e-07,
+      "loss": 2.3913,
+      "step": 7218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.788369688489225e-07,
+      "loss": 2.4025,
+      "step": 7219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.775975995535386e-07,
+      "loss": 2.503,
+      "step": 7220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7635979702158385e-07,
+      "loss": 2.5168,
+      "step": 7221
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7512356145671734e-07,
+      "loss": 2.4596,
+      "step": 7222
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7388889306234197e-07,
+      "loss": 2.5168,
+      "step": 7223
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.726557920416042e-07,
+      "loss": 2.5151,
+      "step": 7224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.714242585973916e-07,
+      "loss": 2.4034,
+      "step": 7225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.701942929323322e-07,
+      "loss": 2.5302,
+      "step": 7226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.689658952488019e-07,
+      "loss": 2.5153,
+      "step": 7227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6773906574891003e-07,
+      "loss": 2.4013,
+      "step": 7228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.66513804634513e-07,
+      "loss": 2.4116,
+      "step": 7229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6529011210721175e-07,
+      "loss": 2.4829,
+      "step": 7230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6406798836834187e-07,
+      "loss": 2.4563,
+      "step": 7231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.628474336189881e-07,
+      "loss": 2.3668,
+      "step": 7232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6162844805997085e-07,
+      "loss": 2.4552,
+      "step": 7233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.604110318918575e-07,
+      "loss": 2.5083,
+      "step": 7234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5919518531495346e-07,
+      "loss": 2.4347,
+      "step": 7235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5798090852931077e-07,
+      "loss": 2.5074,
+      "step": 7236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.56768201734713e-07,
+      "loss": 2.3942,
+      "step": 7237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5555706513069954e-07,
+      "loss": 2.4285,
+      "step": 7238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5434749891653975e-07,
+      "loss": 2.5252,
+      "step": 7239
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5313950329125e-07,
+      "loss": 2.4691,
+      "step": 7240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5193307845358804e-07,
+      "loss": 2.4363,
+      "step": 7241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5072822460205167e-07,
+      "loss": 2.4088,
+      "step": 7242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4952494193488016e-07,
+      "loss": 2.5328,
+      "step": 7243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4832323065005734e-07,
+      "loss": 2.419,
+      "step": 7244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4712309094530283e-07,
+      "loss": 2.4464,
+      "step": 7245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.45924523018082e-07,
+      "loss": 2.4478,
+      "step": 7246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.447275270656037e-07,
+      "loss": 2.4292,
+      "step": 7247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4353210328481055e-07,
+      "loss": 2.4875,
+      "step": 7248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.42338251872394e-07,
+      "loss": 2.5007,
+      "step": 7249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4114597302478244e-07,
+      "loss": 2.5031,
+      "step": 7250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.399552669381468e-07,
+      "loss": 2.4011,
+      "step": 7251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3876613380839925e-07,
+      "loss": 2.4961,
+      "step": 7252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.375785738311955e-07,
+      "loss": 2.3444,
+      "step": 7253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.363925872019259e-07,
+      "loss": 2.5323,
+      "step": 7254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.35208174115731e-07,
+      "loss": 2.412,
+      "step": 7255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.340253347674828e-07,
+      "loss": 2.4989,
+      "step": 7256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.328440693518021e-07,
+      "loss": 2.411,
+      "step": 7257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.316643780630458e-07,
+      "loss": 2.5753,
+      "step": 7258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3048626109531643e-07,
+      "loss": 2.4724,
+      "step": 7259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.293097186424522e-07,
+      "loss": 2.4498,
+      "step": 7260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2813475089803824e-07,
+      "loss": 2.4583,
+      "step": 7261
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.269613580553922e-07,
+      "loss": 2.4276,
+      "step": 7262
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.257895403075818e-07,
+      "loss": 2.4867,
+      "step": 7263
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.246192978474117e-07,
+      "loss": 2.528,
+      "step": 7264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.234506308674236e-07,
+      "loss": 2.4932,
+      "step": 7265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2228353955990563e-07,
+      "loss": 2.4005,
+      "step": 7266
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.211180241168855e-07,
+      "loss": 2.4867,
+      "step": 7267
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.199540847301287e-07,
+      "loss": 2.4338,
+      "step": 7268
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1879172159114414e-07,
+      "loss": 2.449,
+      "step": 7269
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.176309348911822e-07,
+      "loss": 2.4839,
+      "step": 7270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.16471724821228e-07,
+      "loss": 2.4784,
+      "step": 7271
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1531409157201776e-07,
+      "loss": 2.4273,
+      "step": 7272
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1415803533401687e-07,
+      "loss": 2.4761,
+      "step": 7273
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1300355629743884e-07,
+      "loss": 2.3972,
+      "step": 7274
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1185065465223386e-07,
+      "loss": 2.532,
+      "step": 7275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1069933058809577e-07,
+      "loss": 2.4606,
+      "step": 7276
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.095495842944552e-07,
+      "loss": 2.4025,
+      "step": 7277
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0840141596048764e-07,
+      "loss": 2.4404,
+      "step": 7278
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.072548257751019e-07,
+      "loss": 2.4697,
+      "step": 7279
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0610981392695594e-07,
+      "loss": 2.3311,
+      "step": 7280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.049663806044435e-07,
+      "loss": 2.4468,
+      "step": 7281
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.038245259956952e-07,
+      "loss": 2.3708,
+      "step": 7282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.026842502885886e-07,
+      "loss": 2.4719,
+      "step": 7283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0154555367073686e-07,
+      "loss": 2.506,
+      "step": 7284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.004084363294958e-07,
+      "loss": 2.4851,
+      "step": 7285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9927289845196003e-07,
+      "loss": 2.4116,
+      "step": 7286
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.981389402249647e-07,
+      "loss": 2.4547,
+      "step": 7287
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.970065618350838e-07,
+      "loss": 2.4011,
+      "step": 7288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.958757634686361e-07,
+      "loss": 2.4871,
+      "step": 7289
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.947465453116728e-07,
+      "loss": 2.4239,
+      "step": 7290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.936189075499919e-07,
+      "loss": 2.4255,
+      "step": 7291
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.924928503691272e-07,
+      "loss": 2.3937,
+      "step": 7292
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.913683739543539e-07,
+      "loss": 2.5132,
+      "step": 7293
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.902454784906884e-07,
+      "loss": 2.5328,
+      "step": 7294
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8912416416288623e-07,
+      "loss": 2.4885,
+      "step": 7295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.880044311554387e-07,
+      "loss": 2.4567,
+      "step": 7296
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.868862796525841e-07,
+      "loss": 2.4693,
+      "step": 7297
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.857697098382962e-07,
+      "loss": 2.4629,
+      "step": 7298
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8465472189628814e-07,
+      "loss": 2.4107,
+      "step": 7299
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.835413160100132e-07,
+      "loss": 2.3651,
+      "step": 7300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8242949236266814e-07,
+      "loss": 2.5429,
+      "step": 7301
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8131925113718324e-07,
+      "loss": 2.3309,
+      "step": 7302
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.802105925162325e-07,
+      "loss": 2.4927,
+      "step": 7303
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7910351668222987e-07,
+      "loss": 2.4756,
+      "step": 7304
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.779980238173242e-07,
+      "loss": 2.5372,
+      "step": 7305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7689411410340994e-07,
+      "loss": 2.4682,
+      "step": 7306
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7579178772211843e-07,
+      "loss": 2.3987,
+      "step": 7307
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.74691044854818e-07,
+      "loss": 2.4725,
+      "step": 7308
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7359188568262037e-07,
+      "loss": 2.3419,
+      "step": 7309
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.724943103863754e-07,
+      "loss": 2.4123,
+      "step": 7310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.713983191466708e-07,
+      "loss": 2.5055,
+      "step": 7311
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7030391214383787e-07,
+      "loss": 2.4949,
+      "step": 7312
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6921108955793927e-07,
+      "loss": 2.4647,
+      "step": 7313
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.681198515687856e-07,
+      "loss": 2.4183,
+      "step": 7314
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6703019835592436e-07,
+      "loss": 2.4219,
+      "step": 7315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.659421300986377e-07,
+      "loss": 2.342,
+      "step": 7316
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.648556469759501e-07,
+      "loss": 2.496,
+      "step": 7317
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.637707491666298e-07,
+      "loss": 2.4796,
+      "step": 7318
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.626874368491762e-07,
+      "loss": 2.4651,
+      "step": 7319
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.616057102018311e-07,
+      "loss": 2.4268,
+      "step": 7320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6052556940257886e-07,
+      "loss": 2.4627,
+      "step": 7321
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5944701462913735e-07,
+      "loss": 2.4855,
+      "step": 7322
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5837004605896785e-07,
+      "loss": 2.4439,
+      "step": 7323
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.572946638692676e-07,
+      "loss": 2.5111,
+      "step": 7324
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.562208682369739e-07,
+      "loss": 2.469,
+      "step": 7325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.551486593387632e-07,
+      "loss": 2.4233,
+      "step": 7326
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5407803735105215e-07,
+      "loss": 2.4737,
+      "step": 7327
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5300900244999323e-07,
+      "loss": 2.42,
+      "step": 7328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5194155481148127e-07,
+      "loss": 2.3494,
+      "step": 7329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.508756946111447e-07,
+      "loss": 2.4594,
+      "step": 7330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.498114220243587e-07,
+      "loss": 2.4088,
+      "step": 7331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4874873722623103e-07,
+      "loss": 2.433,
+      "step": 7332
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476876403916096e-07,
+      "loss": 2.4293,
+      "step": 7333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4662813169508034e-07,
+      "loss": 2.4987,
+      "step": 7334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.455702113109716e-07,
+      "loss": 2.4743,
+      "step": 7335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4451387941334625e-07,
+      "loss": 2.5079,
+      "step": 7336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.434591361760076e-07,
+      "loss": 2.4341,
+      "step": 7337
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4240598177249673e-07,
+      "loss": 2.5089,
+      "step": 7338
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.41354416376094e-07,
+      "loss": 2.4492,
+      "step": 7339
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4030444015981766e-07,
+      "loss": 2.4449,
+      "step": 7340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.392560532964284e-07,
+      "loss": 2.483,
+      "step": 7341
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3820925595841715e-07,
+      "loss": 2.3692,
+      "step": 7342
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3716404831802053e-07,
+      "loss": 2.5108,
+      "step": 7343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.36120430547211e-07,
+      "loss": 2.5516,
+      "step": 7344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.35078402817699e-07,
+      "loss": 2.4412,
+      "step": 7345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3403796530093733e-07,
+      "loss": 2.5115,
+      "step": 7346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3299911816810804e-07,
+      "loss": 2.5083,
+      "step": 7347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.31961861590141e-07,
+      "loss": 2.4315,
+      "step": 7348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.309261957377019e-07,
+      "loss": 2.49,
+      "step": 7349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2989212078119e-07,
+      "loss": 2.4156,
+      "step": 7350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2885963689074704e-07,
+      "loss": 2.4682,
+      "step": 7351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.27828744236256e-07,
+      "loss": 2.4692,
+      "step": 7352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.267994429873289e-07,
+      "loss": 2.4395,
+      "step": 7353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2577173331332476e-07,
+      "loss": 2.4575,
+      "step": 7354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2474561538333616e-07,
+      "loss": 2.3646,
+      "step": 7355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.237210893661946e-07,
+      "loss": 2.417,
+      "step": 7356
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.226981554304709e-07,
+      "loss": 2.5302,
+      "step": 7357
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.216768137444726e-07,
+      "loss": 2.4252,
+      "step": 7358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.20657064476243e-07,
+      "loss": 2.4333,
+      "step": 7359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.196389077935713e-07,
+      "loss": 2.4901,
+      "step": 7360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.186223438639746e-07,
+      "loss": 2.5121,
+      "step": 7361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.176073728547147e-07,
+      "loss": 2.3966,
+      "step": 7362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.16593994932789e-07,
+      "loss": 2.5082,
+      "step": 7363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1558221026493305e-07,
+      "loss": 2.4583,
+      "step": 7364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.145720190176216e-07,
+      "loss": 2.4257,
+      "step": 7365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1356342135706486e-07,
+      "loss": 2.4405,
+      "step": 7366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.125564174492113e-07,
+      "loss": 2.4263,
+      "step": 7367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.115510074597461e-07,
+      "loss": 2.3794,
+      "step": 7368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1054719155409916e-07,
+      "loss": 2.487,
+      "step": 7369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.095449698974284e-07,
+      "loss": 2.4325,
+      "step": 7370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.085443426546353e-07,
+      "loss": 2.5377,
+      "step": 7371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.07545309990358e-07,
+      "loss": 2.4152,
+      "step": 7372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.065478720689696e-07,
+      "loss": 2.4491,
+      "step": 7373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.055520290545855e-07,
+      "loss": 2.4391,
+      "step": 7374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0455778111105673e-07,
+      "loss": 2.4975,
+      "step": 7375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.035651284019681e-07,
+      "loss": 2.4639,
+      "step": 7376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0257407109064776e-07,
+      "loss": 2.4431,
+      "step": 7377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0158460934015755e-07,
+      "loss": 2.4231,
+      "step": 7378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0059674331329946e-07,
+      "loss": 2.4434,
+      "step": 7379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9961047317260905e-07,
+      "loss": 2.4411,
+      "step": 7380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.986257990803643e-07,
+      "loss": 2.4859,
+      "step": 7381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9764272119857664e-07,
+      "loss": 2.4847,
+      "step": 7382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9666123968899783e-07,
+      "loss": 2.4148,
+      "step": 7383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9568135471311434e-07,
+      "loss": 2.4498,
+      "step": 7384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.947030664321482e-07,
+      "loss": 2.4578,
+      "step": 7385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.937263750070674e-07,
+      "loss": 2.3714,
+      "step": 7386
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.927512805985666e-07,
+      "loss": 2.4401,
+      "step": 7387
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9177778336708427e-07,
+      "loss": 2.4698,
+      "step": 7388
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9080588347279335e-07,
+      "loss": 2.4606,
+      "step": 7389
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8983558107560597e-07,
+      "loss": 2.4461,
+      "step": 7390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8886687633517095e-07,
+      "loss": 2.4802,
+      "step": 7391
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8789976941087206e-07,
+      "loss": 2.416,
+      "step": 7392
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8693426046183083e-07,
+      "loss": 2.4252,
+      "step": 7393
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8597034964691126e-07,
+      "loss": 2.2499,
+      "step": 7394
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.850080371247044e-07,
+      "loss": 2.4044,
+      "step": 7395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8404732305354676e-07,
+      "loss": 2.468,
+      "step": 7396
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8308820759150866e-07,
+      "loss": 2.5582,
+      "step": 7397
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.821306908963972e-07,
+      "loss": 2.503,
+      "step": 7398
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8117477312575636e-07,
+      "loss": 2.4646,
+      "step": 7399
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.802204544368703e-07,
+      "loss": 2.4798,
+      "step": 7400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7926773498675455e-07,
+      "loss": 2.4163,
+      "step": 7401
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7831661493216364e-07,
+      "loss": 2.4432,
+      "step": 7402
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7736709442959255e-07,
+      "loss": 2.4587,
+      "step": 7403
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7641917363526836e-07,
+      "loss": 2.5025,
+      "step": 7404
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7547285270515645e-07,
+      "loss": 2.4754,
+      "step": 7405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7452813179495996e-07,
+      "loss": 2.5046,
+      "step": 7406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.735850110601179e-07,
+      "loss": 2.4142,
+      "step": 7407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.726434906558062e-07,
+      "loss": 2.5109,
+      "step": 7408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7170357073693863e-07,
+      "loss": 2.5116,
+      "step": 7409
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7076525145816045e-07,
+      "loss": 2.4721,
+      "step": 7410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.698285329738626e-07,
+      "loss": 2.4083,
+      "step": 7411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.688934154381639e-07,
+      "loss": 2.3916,
+      "step": 7412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.679598990049248e-07,
+      "loss": 2.4231,
+      "step": 7413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.670279838277401e-07,
+      "loss": 2.4193,
+      "step": 7414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6609767005994267e-07,
+      "loss": 2.3652,
+      "step": 7415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.651689578546024e-07,
+      "loss": 2.4603,
+      "step": 7416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.642418473645236e-07,
+      "loss": 2.4155,
+      "step": 7417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.633163387422455e-07,
+      "loss": 2.4451,
+      "step": 7418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.623924321400495e-07,
+      "loss": 2.5001,
+      "step": 7419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.614701277099485e-07,
+      "loss": 2.479,
+      "step": 7420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6054942560369443e-07,
+      "loss": 2.4068,
+      "step": 7421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5963032597277283e-07,
+      "loss": 2.4963,
+      "step": 7422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.587128289684082e-07,
+      "loss": 2.5299,
+      "step": 7423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5779693474155986e-07,
+      "loss": 2.466,
+      "step": 7424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5688264344292614e-07,
+      "loss": 2.4637,
+      "step": 7425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.559699552229378e-07,
+      "loss": 2.4689,
+      "step": 7426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5505887023176133e-07,
+      "loss": 2.4474,
+      "step": 7427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5414938861930585e-07,
+      "loss": 2.432,
+      "step": 7428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.532415105352104e-07,
+      "loss": 2.3968,
+      "step": 7429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5233523612885115e-07,
+      "loss": 2.4565,
+      "step": 7430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5143056554934317e-07,
+      "loss": 2.4782,
+      "step": 7431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.505274989455353e-07,
+      "loss": 2.5125,
+      "step": 7432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4962603646601414e-07,
+      "loss": 2.3853,
+      "step": 7433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4872617825909995e-07,
+      "loss": 2.4783,
+      "step": 7434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4782792447284985e-07,
+      "loss": 2.4214,
+      "step": 7435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.46931275255059e-07,
+      "loss": 2.4214,
+      "step": 7436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4603623075325823e-07,
+      "loss": 2.4096,
+      "step": 7437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4514279111470975e-07,
+      "loss": 2.4756,
+      "step": 7438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4425095648641815e-07,
+      "loss": 2.4214,
+      "step": 7439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.433607270151195e-07,
+      "loss": 2.4491,
+      "step": 7440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.424721028472876e-07,
+      "loss": 2.4657,
+      "step": 7441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4158508412913117e-07,
+      "loss": 2.4685,
+      "step": 7442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4069967100659786e-07,
+      "loss": 2.4749,
+      "step": 7443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3981586362536444e-07,
+      "loss": 2.4228,
+      "step": 7444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3893366213085243e-07,
+      "loss": 2.4687,
+      "step": 7445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3805306666821015e-07,
+      "loss": 2.5331,
+      "step": 7446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3717407738232835e-07,
+      "loss": 2.4572,
+      "step": 7447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.362966944178302e-07,
+      "loss": 2.4571,
+      "step": 7448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3542091791907584e-07,
+      "loss": 2.5213,
+      "step": 7449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.345467480301611e-07,
+      "loss": 2.4149,
+      "step": 7450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3367418489491645e-07,
+      "loss": 2.3981,
+      "step": 7451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.328032286569071e-07,
+      "loss": 2.4545,
+      "step": 7452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3193387945943835e-07,
+      "loss": 2.4949,
+      "step": 7453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3106613744554696e-07,
+      "loss": 2.4106,
+      "step": 7454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3020000275800535e-07,
+      "loss": 2.417,
+      "step": 7455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2933547553932288e-07,
+      "loss": 2.3302,
+      "step": 7456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2847255593174578e-07,
+      "loss": 2.4709,
+      "step": 7457
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2761124407725266e-07,
+      "loss": 2.4398,
+      "step": 7458
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2675154011755795e-07,
+      "loss": 2.3828,
+      "step": 7459
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.258934441941163e-07,
+      "loss": 2.4983,
+      "step": 7460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2503695644810807e-07,
+      "loss": 2.4499,
+      "step": 7461
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.241820770204617e-07,
+      "loss": 2.4639,
+      "step": 7462
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2332880605182906e-07,
+      "loss": 2.4839,
+      "step": 7463
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2247714368260564e-07,
+      "loss": 2.532,
+      "step": 7464
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2162709005291826e-07,
+      "loss": 2.3945,
+      "step": 7465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.207786453026295e-07,
+      "loss": 2.4489,
+      "step": 7466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1993180957133874e-07,
+      "loss": 2.4615,
+      "step": 7467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1908658299838016e-07,
+      "loss": 2.4094,
+      "step": 7468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1824296572281912e-07,
+      "loss": 2.4298,
+      "step": 7469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.174009578834646e-07,
+      "loss": 2.482,
+      "step": 7470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.165605596188536e-07,
+      "loss": 2.5374,
+      "step": 7471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.15721771067261e-07,
+      "loss": 2.4045,
+      "step": 7472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1488459236669423e-07,
+      "loss": 2.4534,
+      "step": 7473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.14049023654902e-07,
+      "loss": 2.498,
+      "step": 7474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1321506506936206e-07,
+      "loss": 2.4614,
+      "step": 7475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1238271674728805e-07,
+      "loss": 2.3816,
+      "step": 7476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1155197882563262e-07,
+      "loss": 2.4521,
+      "step": 7477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1072285144107863e-07,
+      "loss": 2.4293,
+      "step": 7478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0989533473004807e-07,
+      "loss": 2.4864,
+      "step": 7479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.090694288286932e-07,
+      "loss": 2.4821,
+      "step": 7480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0824513387290634e-07,
+      "loss": 2.4581,
+      "step": 7481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.074224499983124e-07,
+      "loss": 2.4261,
+      "step": 7482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0660137734026976e-07,
+      "loss": 2.4614,
+      "step": 7483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0578191603387478e-07,
+      "loss": 2.4433,
+      "step": 7484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0496406621395627e-07,
+      "loss": 2.4848,
+      "step": 7485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.041478280150777e-07,
+      "loss": 2.452,
+      "step": 7486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0333320157154056e-07,
+      "loss": 2.5676,
+      "step": 7487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0252018701737763e-07,
+      "loss": 2.3748,
+      "step": 7488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0170878448635854e-07,
+      "loss": 2.4474,
+      "step": 7489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0089899411198433e-07,
+      "loss": 2.4554,
+      "step": 7490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0009081602749835e-07,
+      "loss": 2.4309,
+      "step": 7491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9928425036586985e-07,
+      "loss": 2.4746,
+      "step": 7492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9847929725980709e-07,
+      "loss": 2.4387,
+      "step": 7493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.976759568417541e-07,
+      "loss": 2.4011,
+      "step": 7494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9687422924388744e-07,
+      "loss": 2.4447,
+      "step": 7495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9607411459811932e-07,
+      "loss": 2.4336,
+      "step": 7496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9527561303609555e-07,
+      "loss": 2.3413,
+      "step": 7497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.944787246891966e-07,
+      "loss": 2.4414,
+      "step": 7498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.936834496885387e-07,
+      "loss": 2.4812,
+      "step": 7499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9288978816497272e-07,
+      "loss": 2.4462,
+      "step": 7500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9209774024908202e-07,
+      "loss": 2.4933,
+      "step": 7501
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9130730607118786e-07,
+      "loss": 2.4928,
+      "step": 7502
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9051848576133958e-07,
+      "loss": 2.5195,
+      "step": 7503
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8973127944933001e-07,
+      "loss": 2.5379,
+      "step": 7504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.889456872646811e-07,
+      "loss": 2.4824,
+      "step": 7505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8816170933664723e-07,
+      "loss": 2.4103,
+      "step": 7506
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.873793457942208e-07,
+      "loss": 2.4136,
+      "step": 7507
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.865985967661299e-07,
+      "loss": 2.3956,
+      "step": 7508
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8581946238083182e-07,
+      "loss": 2.4722,
+      "step": 7509
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8504194276652178e-07,
+      "loss": 2.5131,
+      "step": 7510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8426603805112852e-07,
+      "loss": 2.4078,
+      "step": 7511
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8349174836231554e-07,
+      "loss": 2.5118,
+      "step": 7512
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8271907382747978e-07,
+      "loss": 2.4626,
+      "step": 7513
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.81948014573754e-07,
+      "loss": 2.3931,
+      "step": 7514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8117857072800117e-07,
+      "loss": 2.4023,
+      "step": 7515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8041074241682555e-07,
+      "loss": 2.4658,
+      "step": 7516
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7964452976655834e-07,
+      "loss": 2.4608,
+      "step": 7517
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7887993290326756e-07,
+      "loss": 2.5267,
+      "step": 7518
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7811695195275924e-07,
+      "loss": 2.4129,
+      "step": 7519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.773555870405641e-07,
+      "loss": 2.4818,
+      "step": 7520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.765958382919586e-07,
+      "loss": 2.4304,
+      "step": 7521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7583770583194602e-07,
+      "loss": 2.3726,
+      "step": 7522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7508118978526334e-07,
+      "loss": 2.3772,
+      "step": 7523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7432629027638425e-07,
+      "loss": 2.3964,
+      "step": 7524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7357300742951832e-07,
+      "loss": 2.3932,
+      "step": 7525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7282134136860418e-07,
+      "loss": 2.428,
+      "step": 7526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.720712922173162e-07,
+      "loss": 2.4976,
+      "step": 7527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7132286009906574e-07,
+      "loss": 2.4295,
+      "step": 7528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.705760451369931e-07,
+      "loss": 2.436,
+      "step": 7529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.698308474539767e-07,
+      "loss": 2.5255,
+      "step": 7530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6908726717262735e-07,
+      "loss": 2.3656,
+      "step": 7531
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6834530441528719e-07,
+      "loss": 2.3902,
+      "step": 7532
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6760495930403963e-07,
+      "loss": 2.4907,
+      "step": 7533
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6686623196069174e-07,
+      "loss": 2.3818,
+      "step": 7534
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6612912250679181e-07,
+      "loss": 2.5075,
+      "step": 7535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.653936310636206e-07,
+      "loss": 2.4017,
+      "step": 7536
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6465975775219023e-07,
+      "loss": 2.4961,
+      "step": 7537
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6392750269324965e-07,
+      "loss": 2.3685,
+      "step": 7538
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6319686600727913e-07,
+      "loss": 2.3296,
+      "step": 7539
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6246784781449364e-07,
+      "loss": 2.481,
+      "step": 7540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6174044823484057e-07,
+      "loss": 2.367,
+      "step": 7541
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610146673880053e-07,
+      "loss": 2.4321,
+      "step": 7542
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.602905053934023e-07,
+      "loss": 2.4758,
+      "step": 7543
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.595679623701796e-07,
+      "loss": 2.4305,
+      "step": 7544
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5884703843722205e-07,
+      "loss": 2.4371,
+      "step": 7545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5812773371314704e-07,
+      "loss": 2.3802,
+      "step": 7546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5741004831630325e-07,
+      "loss": 2.4701,
+      "step": 7547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5669398236477617e-07,
+      "loss": 2.5164,
+      "step": 7548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.559795359763816e-07,
+      "loss": 2.4473,
+      "step": 7549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5526670926867436e-07,
+      "loss": 2.5365,
+      "step": 7550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5455550235893401e-07,
+      "loss": 2.4986,
+      "step": 7551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5384591536418136e-07,
+      "loss": 2.4076,
+      "step": 7552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5313794840116857e-07,
+      "loss": 2.3826,
+      "step": 7553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5243160158637805e-07,
+      "loss": 2.4689,
+      "step": 7554
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5172687503603012e-07,
+      "loss": 2.4869,
+      "step": 7555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.510237688660776e-07,
+      "loss": 2.4337,
+      "step": 7556
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5032228319220354e-07,
+      "loss": 2.4489,
+      "step": 7557
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4962241812982558e-07,
+      "loss": 2.5271,
+      "step": 7558
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4892417379410052e-07,
+      "loss": 2.3954,
+      "step": 7559
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4822755029990864e-07,
+      "loss": 2.5119,
+      "step": 7560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4753254776186942e-07,
+      "loss": 2.3156,
+      "step": 7561
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.468391662943369e-07,
+      "loss": 2.4384,
+      "step": 7562
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4614740601139432e-07,
+      "loss": 2.4469,
+      "step": 7563
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4545726702686168e-07,
+      "loss": 2.4691,
+      "step": 7564
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4476874945428932e-07,
+      "loss": 2.4065,
+      "step": 7565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.44081853406961e-07,
+      "loss": 2.509,
+      "step": 7566
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.433965789978986e-07,
+      "loss": 2.4327,
+      "step": 7567
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4271292633984967e-07,
+      "loss": 2.5103,
+      "step": 7568
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.420308955453009e-07,
+      "loss": 2.4019,
+      "step": 7569
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4135048672646813e-07,
+      "loss": 2.5205,
+      "step": 7570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4067169999530283e-07,
+      "loss": 2.4379,
+      "step": 7571
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3999453546348905e-07,
+      "loss": 2.521,
+      "step": 7572
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3931899324244437e-07,
+      "loss": 2.4231,
+      "step": 7573
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3864507344331646e-07,
+      "loss": 2.3886,
+      "step": 7574
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3797277617699113e-07,
+      "loss": 2.3767,
+      "step": 7575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3730210155408318e-07,
+      "loss": 2.4554,
+      "step": 7576
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3663304968494107e-07,
+      "loss": 2.4223,
+      "step": 7577
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3596562067964781e-07,
+      "loss": 2.4813,
+      "step": 7578
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.352998146480189e-07,
+      "loss": 2.4458,
+      "step": 7579
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3463563169960003e-07,
+      "loss": 2.5568,
+      "step": 7580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3397307194367493e-07,
+      "loss": 2.3603,
+      "step": 7581
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.333121354892575e-07,
+      "loss": 2.4688,
+      "step": 7582
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.326528224450907e-07,
+      "loss": 2.4799,
+      "step": 7583
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3199513291965894e-07,
+      "loss": 2.4783,
+      "step": 7584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3133906702117338e-07,
+      "loss": 2.458,
+      "step": 7585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.306846248575777e-07,
+      "loss": 2.528,
+      "step": 7586
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3003180653655135e-07,
+      "loss": 2.48,
+      "step": 7587
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2938061216550612e-07,
+      "loss": 2.4785,
+      "step": 7588
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2873104185158636e-07,
+      "loss": 2.5189,
+      "step": 7589
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2808309570166766e-07,
+      "loss": 2.5109,
+      "step": 7590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2743677382235809e-07,
+      "loss": 2.4357,
+      "step": 7591
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2679207632000145e-07,
+      "loss": 2.4457,
+      "step": 7592
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2614900330067513e-07,
+      "loss": 2.4147,
+      "step": 7593
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.255075548701823e-07,
+      "loss": 2.3891,
+      "step": 7594
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2486773113406624e-07,
+      "loss": 2.5344,
+      "step": 7595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2422953219759838e-07,
+      "loss": 2.4009,
+      "step": 7596
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2359295816578576e-07,
+      "loss": 2.5298,
+      "step": 7597
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2295800914336574e-07,
+      "loss": 2.4597,
+      "step": 7598
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2232468523481146e-07,
+      "loss": 2.4291,
+      "step": 7599
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2169298654432283e-07,
+      "loss": 2.3819,
+      "step": 7600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2106291317584007e-07,
+      "loss": 2.5656,
+      "step": 7601
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2043446523302916e-07,
+      "loss": 2.4185,
+      "step": 7602
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.198076428192918e-07,
+      "loss": 2.4275,
+      "step": 7603
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.191824460377633e-07,
+      "loss": 2.5065,
+      "step": 7604
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1855887499130803e-07,
+      "loss": 2.436,
+      "step": 7605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1793692978252725e-07,
+      "loss": 2.3814,
+      "step": 7606
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1731661051375021e-07,
+      "loss": 2.4855,
+      "step": 7607
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1669791728704194e-07,
+      "loss": 2.4487,
+      "step": 7608
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1608085020419879e-07,
+      "loss": 2.3579,
+      "step": 7609
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1546540936675066e-07,
+      "loss": 2.4405,
+      "step": 7610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1485159487595654e-07,
+      "loss": 2.401,
+      "step": 7611
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1423940683281121e-07,
+      "loss": 2.4115,
+      "step": 7612
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1362884533803964e-07,
+      "loss": 2.4608,
+      "step": 7613
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1301991049210259e-07,
+      "loss": 2.4699,
+      "step": 7614
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1241260239518881e-07,
+      "loss": 2.4037,
+      "step": 7615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.118069211472228e-07,
+      "loss": 2.4556,
+      "step": 7616
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1120286684785708e-07,
+      "loss": 2.3656,
+      "step": 7617
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1060043959648436e-07,
+      "loss": 2.4196,
+      "step": 7618
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0999963949221981e-07,
+      "loss": 2.4483,
+      "step": 7619
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0940046663391879e-07,
+      "loss": 2.4627,
+      "step": 7620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0880292112016466e-07,
+      "loss": 2.4304,
+      "step": 7621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0820700304927434e-07,
+      "loss": 2.3627,
+      "step": 7622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0761271251929717e-07,
+      "loss": 2.4156,
+      "step": 7623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0702004962801493e-07,
+      "loss": 2.41,
+      "step": 7624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0642901447293963e-07,
+      "loss": 2.4202,
+      "step": 7625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0583960715131791e-07,
+      "loss": 2.4375,
+      "step": 7626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0525182776012888e-07,
+      "loss": 2.3855,
+      "step": 7627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.046656763960796e-07,
+      "loss": 2.5576,
+      "step": 7628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0408115315561406e-07,
+      "loss": 2.4232,
+      "step": 7629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0349825813490644e-07,
+      "loss": 2.4427,
+      "step": 7630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0291699142986222e-07,
+      "loss": 2.4356,
+      "step": 7631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0233735313612048e-07,
+      "loss": 2.3991,
+      "step": 7632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0175934334905158e-07,
+      "loss": 2.3372,
+      "step": 7633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0118296216375833e-07,
+      "loss": 2.5259,
+      "step": 7634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0060820967507601e-07,
+      "loss": 2.5412,
+      "step": 7635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0003508597756783e-07,
+      "loss": 2.4757,
+      "step": 7636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.946359116553616e-08,
+      "loss": 2.4581,
+      "step": 7637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.889372533301023e-08,
+      "loss": 2.4313,
+      "step": 7638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.832548857375168e-08,
+      "loss": 2.3936,
+      "step": 7639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.775888098125686e-08,
+      "loss": 2.5082,
+      "step": 7640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.719390264875117e-08,
+      "loss": 2.3858,
+      "step": 7641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.66305536691925e-08,
+      "loss": 2.4771,
+      "step": 7642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.606883413527224e-08,
+      "loss": 2.5249,
+      "step": 7643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.550874413941314e-08,
+      "loss": 2.455,
+      "step": 7644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.495028377376703e-08,
+      "loss": 2.3884,
+      "step": 7645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.439345313022153e-08,
+      "loss": 2.5262,
+      "step": 7646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.383825230039445e-08,
+      "loss": 2.4549,
+      "step": 7647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.328468137563607e-08,
+      "loss": 2.4639,
+      "step": 7648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.273274044702574e-08,
+      "loss": 2.3364,
+      "step": 7649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.21824296053786e-08,
+      "loss": 2.4498,
+      "step": 7650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.163374894123889e-08,
+      "loss": 2.4453,
+      "step": 7651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.108669854488439e-08,
+      "loss": 2.5248,
+      "step": 7652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.054127850632311e-08,
+      "loss": 2.5057,
+      "step": 7653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.999748891529547e-08,
+      "loss": 2.3987,
+      "step": 7654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.945532986127326e-08,
+      "loss": 2.4028,
+      "step": 7655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.891480143346065e-08,
+      "loss": 2.4724,
+      "step": 7656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.837590372079318e-08,
+      "loss": 2.4794,
+      "step": 7657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.783863681193882e-08,
+      "loss": 2.5613,
+      "step": 7658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.730300079529575e-08,
+      "loss": 2.4703,
+      "step": 7659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.676899575899456e-08,
+      "loss": 2.4884,
+      "step": 7660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.62366217908972e-08,
+      "loss": 2.4135,
+      "step": 7661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.570587897859806e-08,
+      "loss": 2.4886,
+      "step": 7662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.517676740942282e-08,
+      "loss": 2.3943,
+      "step": 7663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.464928717042852e-08,
+      "loss": 2.5227,
+      "step": 7664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.412343834840242e-08,
+      "loss": 2.4799,
+      "step": 7665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.359922102986751e-08,
+      "loss": 2.5552,
+      "step": 7666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.307663530107258e-08,
+      "loss": 2.4787,
+      "step": 7667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.255568124800329e-08,
+      "loss": 2.4867,
+      "step": 7668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.20363589563733e-08,
+      "loss": 2.4504,
+      "step": 7669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.151866851162982e-08,
+      "loss": 2.4243,
+      "step": 7670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.100260999894915e-08,
+      "loss": 2.3848,
+      "step": 7671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.048818350324338e-08,
+      "loss": 2.4594,
+      "step": 7672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.997538910915148e-08,
+      "loss": 2.4905,
+      "step": 7673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.946422690104594e-08,
+      "loss": 2.452,
+      "step": 7674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.89546969630317e-08,
+      "loss": 2.4159,
+      "step": 7675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.844679937894283e-08,
+      "loss": 2.3534,
+      "step": 7676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.794053423234582e-08,
+      "loss": 2.4979,
+      "step": 7677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.743590160654068e-08,
+      "loss": 2.5063,
+      "step": 7678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.693290158455435e-08,
+      "loss": 2.5527,
+      "step": 7679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.643153424914729e-08,
+      "loss": 2.4503,
+      "step": 7680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.593179968281573e-08,
+      "loss": 2.5342,
+      "step": 7681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.543369796777944e-08,
+      "loss": 2.4904,
+      "step": 7682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.493722918599511e-08,
+      "loss": 2.502,
+      "step": 7683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.444239341914849e-08,
+      "loss": 2.3569,
+      "step": 7684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.39491907486567e-08,
+      "loss": 2.4431,
+      "step": 7685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.345762125566924e-08,
+      "loss": 2.3885,
+      "step": 7686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.296768502106589e-08,
+      "loss": 2.5041,
+      "step": 7687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.247938212545657e-08,
+      "loss": 2.4716,
+      "step": 7688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.199271264918706e-08,
+      "loss": 2.4015,
+      "step": 7689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.150767667232994e-08,
+      "loss": 2.3803,
+      "step": 7690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.102427427468916e-08,
+      "loss": 2.4594,
+      "step": 7691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.054250553580222e-08,
+      "loss": 2.3628,
+      "step": 7692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.00623705349368e-08,
+      "loss": 2.5318,
+      "step": 7693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.958386935109196e-08,
+      "loss": 2.4748,
+      "step": 7694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.91070020629958e-08,
+      "loss": 2.4469,
+      "step": 7695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.863176874911116e-08,
+      "loss": 2.3525,
+      "step": 7696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.815816948762988e-08,
+      "loss": 2.4692,
+      "step": 7697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.768620435647522e-08,
+      "loss": 2.4224,
+      "step": 7698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.721587343330172e-08,
+      "loss": 2.4731,
+      "step": 7699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.674717679549414e-08,
+      "loss": 2.4966,
+      "step": 7700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.62801145201708e-08,
+      "loss": 2.4617,
+      "step": 7701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.581468668417912e-08,
+      "loss": 2.3131,
+      "step": 7702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.535089336409784e-08,
+      "loss": 2.436,
+      "step": 7703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.488873463623701e-08,
+      "loss": 2.4516,
+      "step": 7704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.442821057663696e-08,
+      "loss": 2.4442,
+      "step": 7705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.396932126107036e-08,
+      "loss": 2.5516,
+      "step": 7706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.351206676504129e-08,
+      "loss": 2.4033,
+      "step": 7707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.305644716378179e-08,
+      "loss": 2.5599,
+      "step": 7708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.260246253225854e-08,
+      "loss": 2.3542,
+      "step": 7709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.215011294516737e-08,
+      "loss": 2.4272,
+      "step": 7710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.16993984769354e-08,
+      "loss": 2.4193,
+      "step": 7711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.125031920171997e-08,
+      "loss": 2.4965,
+      "step": 7712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.080287519341088e-08,
+      "loss": 2.4503,
+      "step": 7713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.035706652562811e-08,
+      "loss": 2.4148,
+      "step": 7714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9912893271723e-08,
+      "loss": 2.4778,
+      "step": 7715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9470355504774865e-08,
+      "loss": 2.4322,
+      "step": 7716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9029453297599905e-08,
+      "loss": 2.4343,
+      "step": 7717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.859018672273897e-08,
+      "loss": 2.4909,
+      "step": 7718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.815255585246871e-08,
+      "loss": 2.3566,
+      "step": 7719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7716560758792614e-08,
+      "loss": 2.393,
+      "step": 7720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7282201513447766e-08,
+      "loss": 2.4111,
+      "step": 7721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.684947818790143e-08,
+      "loss": 2.4688,
+      "step": 7722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.64183908533511e-08,
+      "loss": 2.4633,
+      "step": 7723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.59889395807256e-08,
+      "loss": 2.4124,
+      "step": 7724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.556112444068507e-08,
+      "loss": 2.4794,
+      "step": 7725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5134945503618754e-08,
+      "loss": 2.4369,
+      "step": 7726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.471040283964835e-08,
+      "loss": 2.4333,
+      "step": 7727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.428749651862686e-08,
+      "loss": 2.4729,
+      "step": 7728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3866226610135295e-08,
+      "loss": 2.4476,
+      "step": 7729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.344659318348599e-08,
+      "loss": 2.4857,
+      "step": 7730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3028596307725946e-08,
+      "loss": 2.484,
+      "step": 7731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.261223605163013e-08,
+      "loss": 2.49,
+      "step": 7732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.219751248370153e-08,
+      "loss": 2.3834,
+      "step": 7733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.178442567217779e-08,
+      "loss": 2.4096,
+      "step": 7734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.137297568502564e-08,
+      "loss": 2.4316,
+      "step": 7735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.096316258994316e-08,
+      "loss": 2.5067,
+      "step": 7736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.055498645435974e-08,
+      "loss": 2.3153,
+      "step": 7737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.014844734543278e-08,
+      "loss": 2.4837,
+      "step": 7738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.97435453300521e-08,
+      "loss": 2.4324,
+      "step": 7739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9340280474838855e-08,
+      "loss": 2.4515,
+      "step": 7740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.89386528461433e-08,
+      "loss": 2.3569,
+      "step": 7741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.853866251004813e-08,
+      "loss": 2.3931,
+      "step": 7742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.814030953236404e-08,
+      "loss": 2.399,
+      "step": 7743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.774359397863526e-08,
+      "loss": 2.4397,
+      "step": 7744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.734851591413403e-08,
+      "loss": 2.5532,
+      "step": 7745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.695507540386501e-08,
+      "loss": 2.4638,
+      "step": 7746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656327251256198e-08,
+      "loss": 2.533,
+      "step": 7747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6173107304690043e-08,
+      "loss": 2.4679,
+      "step": 7748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.578457984444673e-08,
+      "loss": 2.4101,
+      "step": 7749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.539769019575646e-08,
+      "loss": 2.4632,
+      "step": 7750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.501243842227498e-08,
+      "loss": 2.4839,
+      "step": 7751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.462882458739159e-08,
+      "loss": 2.4704,
+      "step": 7752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4246848754223584e-08,
+      "loss": 2.5225,
+      "step": 7753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.386651098561845e-08,
+      "loss": 2.4693,
+      "step": 7754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3487811344155027e-08,
+      "loss": 2.4304,
+      "step": 7755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3110749892142366e-08,
+      "loss": 2.4596,
+      "step": 7756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.273532669162084e-08,
+      "loss": 2.4801,
+      "step": 7757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.236154180435881e-08,
+      "loss": 2.4086,
+      "step": 7758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1989395291858195e-08,
+      "loss": 2.458,
+      "step": 7759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.161888721535001e-08,
+      "loss": 2.5025,
+      "step": 7760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.12500176357955e-08,
+      "loss": 2.5308,
+      "step": 7761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0882786613885005e-08,
+      "loss": 2.4079,
+      "step": 7762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.051719421004352e-08,
+      "loss": 2.5273,
+      "step": 7763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.015324048442071e-08,
+      "loss": 2.5052,
+      "step": 7764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.979092549690089e-08,
+      "loss": 2.4497,
+      "step": 7765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.943024930709749e-08,
+      "loss": 2.4161,
+      "step": 7766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.907121197435304e-08,
+      "loss": 2.4785,
+      "step": 7767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.871381355774251e-08,
+      "loss": 2.3615,
+      "step": 7768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.835805411607108e-08,
+      "loss": 2.5176,
+      "step": 7769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.800393370787081e-08,
+      "loss": 2.4292,
+      "step": 7770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.765145239140955e-08,
+      "loss": 2.5141,
+      "step": 7771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.730061022468201e-08,
+      "loss": 2.4995,
+      "step": 7772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.695140726541202e-08,
+      "loss": 2.3608,
+      "step": 7773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.660384357105806e-08,
+      "loss": 2.4979,
+      "step": 7774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.625791919880328e-08,
+      "loss": 2.4935,
+      "step": 7775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5913634205567705e-08,
+      "loss": 2.4056,
+      "step": 7776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5570988647994906e-08,
+      "loss": 2.4499,
+      "step": 7777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.522998258246424e-08,
+      "loss": 2.4659,
+      "step": 7778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.489061606508082e-08,
+      "loss": 2.4977,
+      "step": 7779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.455288915168442e-08,
+      "loss": 2.4242,
+      "step": 7780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4216801897840605e-08,
+      "loss": 2.5017,
+      "step": 7781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388235435884957e-08,
+      "loss": 2.4523,
+      "step": 7782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3549546589737306e-08,
+      "loss": 2.4878,
+      "step": 7783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3218378645264446e-08,
+      "loss": 2.4863,
+      "step": 7784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.288885057991631e-08,
+      "loss": 2.5263,
+      "step": 7785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.256096244791507e-08,
+      "loss": 2.4348,
+      "step": 7786
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.223471430320646e-08,
+      "loss": 2.4576,
+      "step": 7787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.19101061994731e-08,
+      "loss": 2.4807,
+      "step": 7788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.158713819012116e-08,
+      "loss": 2.5245,
+      "step": 7789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1265810328290345e-08,
+      "loss": 2.479,
+      "step": 7790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.094612266685282e-08,
+      "loss": 2.5118,
+      "step": 7791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.062807525840428e-08,
+      "loss": 2.3142,
+      "step": 7792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0311668155277306e-08,
+      "loss": 2.4281,
+      "step": 7793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9996901409531376e-08,
+      "loss": 2.4829,
+      "step": 7794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9683775072956166e-08,
+      "loss": 2.4953,
+      "step": 7795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9372289197071578e-08,
+      "loss": 2.4229,
+      "step": 7796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9062443833127728e-08,
+      "loss": 2.381,
+      "step": 7797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8754239032104948e-08,
+      "loss": 2.4124,
+      "step": 7798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.844767484471378e-08,
+      "loss": 2.5332,
+      "step": 7799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.814275132139388e-08,
+      "loss": 2.4658,
+      "step": 7800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7839468512317336e-08,
+      "loss": 2.4022,
+      "step": 7801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7537826467382013e-08,
+      "loss": 2.3879,
+      "step": 7802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7237825236221537e-08,
+      "loss": 2.4312,
+      "step": 7803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.69394648681931e-08,
+      "loss": 2.456,
+      "step": 7804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.664274541239076e-08,
+      "loss": 2.418,
+      "step": 7805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6347666917632132e-08,
+      "loss": 2.4168,
+      "step": 7806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6054229432469493e-08,
+      "loss": 2.5141,
+      "step": 7807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.576243300518311e-08,
+      "loss": 2.447,
+      "step": 7808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5472277683783464e-08,
+      "loss": 2.4968,
+      "step": 7809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5183763516010152e-08,
+      "loss": 2.4509,
+      "step": 7810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4896890549335197e-08,
+      "loss": 2.4751,
+      "step": 7811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4611658830958617e-08,
+      "loss": 2.4783,
+      "step": 7812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4328068407810656e-08,
+      "loss": 2.4271,
+      "step": 7813
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4046119326551765e-08,
+      "loss": 2.4408,
+      "step": 7814
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.37658116335715e-08,
+      "loss": 2.4587,
+      "step": 7815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.348714537499186e-08,
+      "loss": 2.5032,
+      "step": 7816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3210120596661723e-08,
+      "loss": 2.406,
+      "step": 7817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2934737344162406e-08,
+      "loss": 2.4374,
+      "step": 7818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.266099566280211e-08,
+      "loss": 2.4189,
+      "step": 7819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2388895597622583e-08,
+      "loss": 2.418,
+      "step": 7820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2118437193392462e-08,
+      "loss": 2.5378,
+      "step": 7821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.18496204946117e-08,
+      "loss": 2.5009,
+      "step": 7822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1582445545510477e-08,
+      "loss": 2.4318,
+      "step": 7823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.131691239004807e-08,
+      "loss": 2.4528,
+      "step": 7824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.105302107191398e-08,
+      "loss": 2.4251,
+      "step": 7825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0790771634526807e-08,
+      "loss": 2.4105,
+      "step": 7826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.053016412103537e-08,
+      "loss": 2.4666,
+      "step": 7827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0271198574319804e-08,
+      "loss": 2.4909,
+      "step": 7828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.001387503698826e-08,
+      "loss": 2.4092,
+      "step": 7829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.975819355137909e-08,
+      "loss": 2.47,
+      "step": 7830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.950415415956086e-08,
+      "loss": 2.4372,
+      "step": 7831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9251756903331254e-08,
+      "loss": 2.4285,
+      "step": 7832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9001001824220378e-08,
+      "loss": 2.4655,
+      "step": 7833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.875188896348412e-08,
+      "loss": 2.5285,
+      "step": 7834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.85044183621097e-08,
+      "loss": 2.4604,
+      "step": 7835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.825859006081676e-08,
+      "loss": 2.4372,
+      "step": 7836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8014404100050732e-08,
+      "loss": 2.4363,
+      "step": 7837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.777186051998947e-08,
+      "loss": 2.5472,
+      "step": 7838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7530959360538836e-08,
+      "loss": 2.3892,
+      "step": 7839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7291700661337117e-08,
+      "loss": 2.4436,
+      "step": 7840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.705408446174839e-08,
+      "loss": 2.4714,
+      "step": 7841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6818110800871366e-08,
+      "loss": 2.4507,
+      "step": 7842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6583779717529445e-08,
+      "loss": 2.4013,
+      "step": 7843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.635109125027845e-08,
+      "loss": 2.4208,
+      "step": 7844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.612004543740442e-08,
+      "loss": 2.3513,
+      "step": 7845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.58906423169225e-08,
+      "loss": 2.4277,
+      "step": 7846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5662881926576946e-08,
+      "loss": 2.2729,
+      "step": 7847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.543676430384222e-08,
+      "loss": 2.4496,
+      "step": 7848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5212289485923016e-08,
+      "loss": 2.4206,
+      "step": 7849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4989457509752004e-08,
+      "loss": 2.4457,
+      "step": 7850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4768268411994302e-08,
+      "loss": 2.4121,
+      "step": 7851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4548722229041912e-08,
+      "loss": 2.5248,
+      "step": 7852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4330818997018159e-08,
+      "loss": 2.4003,
+      "step": 7853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4114558751775475e-08,
+      "loss": 2.4846,
+      "step": 7854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.389994152889651e-08,
+      "loss": 2.367,
+      "step": 7855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3686967363693016e-08,
+      "loss": 2.4881,
+      "step": 7856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3475636291205852e-08,
+      "loss": 2.3895,
+      "step": 7857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3265948346208312e-08,
+      "loss": 2.5495,
+      "step": 7858
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.305790356319836e-08,
+      "loss": 2.4549,
+      "step": 7859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2851501976408609e-08,
+      "loss": 2.4462,
+      "step": 7860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2646743619798563e-08,
+      "loss": 2.411,
+      "step": 7861
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2443628527059048e-08,
+      "loss": 2.3295,
+      "step": 7862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2242156731607779e-08,
+      "loss": 2.4548,
+      "step": 7863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2042328266594905e-08,
+      "loss": 2.4036,
+      "step": 7864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.184414316489857e-08,
+      "loss": 2.5087,
+      "step": 7865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1647601459128243e-08,
+      "loss": 2.4911,
+      "step": 7866
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1452703181620283e-08,
+      "loss": 2.4807,
+      "step": 7867
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1259448364442372e-08,
+      "loss": 2.4917,
+      "step": 7868
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1067837039392403e-08,
+      "loss": 2.4754,
+      "step": 7869
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0877869237996275e-08,
+      "loss": 2.3981,
+      "step": 7870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0689544991511202e-08,
+      "loss": 2.3986,
+      "step": 7871
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0502864330922401e-08,
+      "loss": 2.4837,
+      "step": 7872
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0317827286945303e-08,
+      "loss": 2.4249,
+      "step": 7873
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0134433890024443e-08,
+      "loss": 2.4455,
+      "step": 7874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.95268417033457e-09,
+      "loss": 2.4319,
+      "step": 7875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.772578157780654e-09,
+      "loss": 2.4514,
+      "step": 7876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.594115881995658e-09,
+      "loss": 2.4811,
+      "step": 7877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.41729737234276e-09,
+      "loss": 2.4403,
+      "step": 7878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.242122657914244e-09,
+      "loss": 2.4861,
+      "step": 7879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.068591767533719e-09,
+      "loss": 2.4149,
+      "step": 7880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.896704729751682e-09,
+      "loss": 2.495,
+      "step": 7881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.726461572849953e-09,
+      "loss": 2.5563,
+      "step": 7882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.557862324838352e-09,
+      "loss": 2.4685,
+      "step": 7883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.390907013460236e-09,
+      "loss": 2.3902,
+      "step": 7884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.22559566618253e-09,
+      "loss": 2.4753,
+      "step": 7885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.061928310205692e-09,
+      "loss": 2.4366,
+      "step": 7886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.899904972459294e-09,
+      "loss": 2.4403,
+      "step": 7887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.739525679600902e-09,
+      "loss": 2.4325,
+      "step": 7888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.580790458018294e-09,
+      "loss": 2.4843,
+      "step": 7889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.42369933382947e-09,
+      "loss": 2.4778,
+      "step": 7890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.268252332880421e-09,
+      "loss": 2.5152,
+      "step": 7891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.114449480749574e-09,
+      "loss": 2.5356,
+      "step": 7892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.962290802740024e-09,
+      "loss": 2.4356,
+      "step": 7893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.811776323889519e-09,
+      "loss": 2.4373,
+      "step": 7894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.662906068961583e-09,
+      "loss": 2.3725,
+      "step": 7895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.515680062451068e-09,
+      "loss": 2.468,
+      "step": 7896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.370098328581931e-09,
+      "loss": 2.4543,
+      "step": 7897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.2261608913061215e-09,
+      "loss": 2.4301,
+      "step": 7898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0838677743069175e-09,
+      "loss": 2.5305,
+      "step": 7899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.943219000997813e-09,
+      "loss": 2.4579,
+      "step": 7900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.804214594518076e-09,
+      "loss": 2.5054,
+      "step": 7901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.666854577740521e-09,
+      "loss": 2.5117,
+      "step": 7902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.531138973264849e-09,
+      "loss": 2.4673,
+      "step": 7903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3970678034209746e-09,
+      "loss": 2.4545,
+      "step": 7904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.264641090269029e-09,
+      "loss": 2.5866,
+      "step": 7905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.13385885559603e-09,
+      "loss": 2.3616,
+      "step": 7906
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.00472112092143e-09,
+      "loss": 2.561,
+      "step": 7907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.877227907492677e-09,
+      "loss": 2.393,
+      "step": 7908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.751379236287435e-09,
+      "loss": 2.4813,
+      "step": 7909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.627175128010253e-09,
+      "loss": 2.4522,
+      "step": 7910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.504615603099227e-09,
+      "loss": 2.4372,
+      "step": 7911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3837006817182285e-09,
+      "loss": 2.5053,
+      "step": 7912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.264430383762452e-09,
+      "loss": 2.6112,
+      "step": 7913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.146804728855092e-09,
+      "loss": 2.4529,
+      "step": 7914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0308237363506645e-09,
+      "loss": 2.4356,
+      "step": 7915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.916487425331683e-09,
+      "loss": 2.3769,
+      "step": 7916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.803795814610877e-09,
+      "loss": 2.4698,
+      "step": 7917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6927489227289725e-09,
+      "loss": 2.467,
+      "step": 7918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.583346767958018e-09,
+      "loss": 2.4338,
+      "step": 7919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.47558936829695e-09,
+      "loss": 2.4924,
+      "step": 7920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.36947674147714e-09,
+      "loss": 2.5638,
+      "step": 7921
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2650089049568456e-09,
+      "loss": 2.4036,
+      "step": 7922
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1621858759256496e-09,
+      "loss": 2.4651,
+      "step": 7923
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0610076713000203e-09,
+      "loss": 2.4314,
+      "step": 7924
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.961474307728862e-09,
+      "loss": 2.4811,
+      "step": 7925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8635858015868545e-09,
+      "loss": 2.405,
+      "step": 7926
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7673421689822234e-09,
+      "loss": 2.463,
+      "step": 7927
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6727434257489693e-09,
+      "loss": 2.4533,
+      "step": 7928
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5797895874524192e-09,
+      "loss": 2.4232,
+      "step": 7929
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4884806693858953e-09,
+      "loss": 2.4487,
+      "step": 7930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.398816686574046e-09,
+      "loss": 2.4623,
+      "step": 7931
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.310797653768404e-09,
+      "loss": 2.5482,
+      "step": 7932
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2244235854518294e-09,
+      "loss": 2.5456,
+      "step": 7933
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1396944958362866e-09,
+      "loss": 2.3671,
+      "step": 7934
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.056610398861736e-09,
+      "loss": 2.5142,
+      "step": 7935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9751713081994637e-09,
+      "loss": 2.5039,
+      "step": 7936
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.895377237247642e-09,
+      "loss": 2.4886,
+      "step": 7937
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.817228199136878e-09,
+      "loss": 2.5099,
+      "step": 7938
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.740724206723554e-09,
+      "loss": 2.4493,
+      "step": 7939
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6658652725953794e-09,
+      "loss": 2.4717,
+      "step": 7940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5926514090713886e-09,
+      "loss": 2.5097,
+      "step": 7941
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5210826281952807e-09,
+      "loss": 2.4192,
+      "step": 7942
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4511589417431915e-09,
+      "loss": 2.484,
+      "step": 7943
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3828803612214724e-09,
+      "loss": 2.4672,
+      "step": 7944
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.31624689786225e-09,
+      "loss": 2.4954,
+      "step": 7945
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2512585626300866e-09,
+      "loss": 2.4594,
+      "step": 7946
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1879153662175403e-09,
+      "loss": 2.4205,
+      "step": 7947
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1262173190473846e-09,
+      "loss": 2.3949,
+      "step": 7948
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.066164431270389e-09,
+      "loss": 2.4146,
+      "step": 7949
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0077567127675382e-09,
+      "loss": 2.4881,
+      "step": 7950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.509941731489225e-10,
+      "loss": 2.5085,
+      "step": 7951
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.958768217537384e-10,
+      "loss": 2.4868,
+      "step": 7952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.424046676502873e-10,
+      "loss": 2.5498,
+      "step": 7953
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.905777196381969e-10,
+      "loss": 2.3164,
+      "step": 7954
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.4039598624287e-10,
+      "loss": 2.4643,
+      "step": 7955
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.918594757221453e-10,
+      "loss": 2.5599,
+      "step": 7956
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.44968196061857e-10,
+      "loss": 2.5123,
+      "step": 7957
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.99722154976945e-10,
+      "loss": 2.4101,
+      "step": 7958
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.561213599114546e-10,
+      "loss": 2.4882,
+      "step": 7959
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.14165818039647e-10,
+      "loss": 2.3844,
+      "step": 7960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.738555362659991e-10,
+      "loss": 2.4441,
+      "step": 7961
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.35190521220763e-10,
+      "loss": 2.3814,
+      "step": 7962
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9817077926662673e-10,
+      "loss": 2.3918,
+      "step": 7963
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.627963164942738e-10,
+      "loss": 2.4603,
+      "step": 7964
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.290671387246036e-10,
+      "loss": 2.4065,
+      "step": 7965
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9698325150651075e-10,
+      "loss": 2.4638,
+      "step": 7966
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6654466012021594e-10,
+      "loss": 2.4564,
+      "step": 7967
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.377513695728251e-10,
+      "loss": 2.3043,
+      "step": 7968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1060338460165974e-10,
+      "loss": 2.4393,
+      "step": 7969
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851007096742574e-10,
+      "loss": 2.4846,
+      "step": 7970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.612433489861509e-10,
+      "loss": 2.4636,
+      "step": 7971
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3903130646308882e-10,
+      "loss": 2.5588,
+      "step": 7972
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1846458575881514e-10,
+      "loss": 2.3954,
+      "step": 7973
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.954319025839987e-11,
+      "loss": 2.3532,
+      "step": 7974
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.226712307490836e-11,
+      "loss": 2.3792,
+      "step": 7975
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.663638705051157e-11,
+      "loss": 2.4431,
+      "step": 7976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.2650984756486e-11,
+      "loss": 2.4702,
+      "step": 7977
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0310918495434226e-11,
+      "loss": 2.4907,
+      "step": 7978
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9616190295733704e-11,
+      "loss": 2.5207,
+      "step": 7979
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0566801918198153e-11,
+      "loss": 2.5171,
+      "step": 7980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.316275485163665e-11,
+      "loss": 2.3872,
+      "step": 7981
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.404050315074075e-12,
+      "loss": 2.44,
+      "step": 7982
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2906892544204426e-12,
+      "loss": 2.3824,
+      "step": 7983
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.226723469118015e-13,
+      "loss": 2.4161,
+      "step": 7984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 2.4266,
+      "step": 7985
+    },
+    {
+      "epoch": 1.0,
+      "step": 7985,
+      "total_flos": 534287571517440.0,
+      "train_loss": 2.5154464183633,
+      "train_runtime": 46400.6677,
+      "train_samples_per_second": 27.531,
+      "train_steps_per_second": 0.172
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 7985,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 534287571517440.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}